Bachelor’s Thesis Bauke Kok 10624317 31-01-2018 Supervisor: Dirk Damsma
BSc Economics and Business University of Amsterdam
Beyond the rules of Rodrik: a critical assessment of Dani Rodrik’s economic methodology as stated in Economics Rules
Abstract
After the 2008 financial crash much criticism has been raised against economics. In his 2015 book Economics Rules, Dani Rodrik defends the legitimacy of economics by arguing that this criticism is mostly misguided. He presents an image of economics as a practice of carefully applying models to the world in order to identify mechanisms. He calls for modesty in economics and argues that economists should refrain from grand-scale theorizing. This appeal to modesty is in the end unconvincing. Economists cannot take into account the structures that cause the mechanisms they analyse to exist, without looking beyond Rodrik’s account of economics. Moreover, Rodrik’s account offers a narrow normative view of what the economy should be, which is a problem given the role economists play in shaping the world.
2 Contents
1. Introduction 3
2. Economics Rules 4
2.1 In defense of mathematical models 4 2.2 The strength of mathematical models 7
2.3 Using models 8
2.4 The rights and wrongs of economics 9
2.5 The rules of economics 11
3. A structured assessment 11
3.1 Methodological critique 12
3.2 Real world critique 13
3.3 Methodological defense 13
3.3.1 Models 14
3.3.2 Economic ontology 15
3.3.3 Mechanisms 16
3.4 Real world defense 18
3.4.1 Models as blueprints for socio-economic machines 19
3.4.2 The performativity thesis 20
3.4.3 Growth diagnostics 21
4. Conclusion: too much modesty? 21
Bibliography 22
This document is written by Student Bauke Kok who declares to take full responsibility for the contents of this document.
I declare that the text and the work presented in this document are original and that no sources other than those mentioned in the text and its references have been used in creating it.
The Faculty of Economics and Business is responsible solely for the supervision of completion of the work, not for the contents.
3 1. Introduction
The subject of this thesis can be traced back the 2008 financial crisis and the effects of its aftermath on the profession of economics. Many commentators were critical of professional economists for failing to prevent, or even predict the crash. In 2010 Joseph Stiglitz wrote in a Financial Times column: “Today, not only is our economy in a shambles but so too is the economic paradigm that predominated in the years before the crisis.”1 The paradigm referred to is that of neoclassical economics, seen as a homogeneous framework of what the economy is and how it should be studied. The validity of this homogeneous view is then considered to be challenged by the 2008 crash, since this was an event that was at the same time caused by neoclassical economic thinking, and incomprehensible within its framework.
The view of the 2008 financial crash as decisive evidence against the legitimacy of orthodox economic science is disputed by Dani Rodrik in his 2015 book Economics Rules. This is not because economists bear no blame, but because the blame they bear can be explained by their lack of a proper understanding of how economics should be practiced and what its limits are. This leaves economic science, as properly understood, intact.
The validity of Rodrik’s argument is of great importance to the profession of economics. Post-crash critique of economics has put pressure on the field to make major changes to both its way of teaching and of doing research. However, if Rodrik’s view is correct, much criticism of economics is poorly informed and professional economists must be seen as capable of solving the problems that the field faces.
In this thesis, I will take a close look at the arguments that Rodrik puts forward. By scrutinizing these, I will try to answer the question whether Rodrik’s defense of economics is fully convincing. First I will present a summary of Economics Rules. Then I will show how the book is structured, by highlighting two key distinctions: that between Rodrik’s defense of proper economics and his critique of bad economics and that between his abstract methodological argument and his reference to real-world examples. Then I will discuss Rodrik’s defense of economics by analyzing it in the light of arguments made by other authors working in philosophy of science and economic methodology. This will allow me to raise some critical points that, I believe, show that Rodrik has not ended the debate on the legitimacy of economics. Instead, there is a need for further methodological theorizing in economics, and further critical discussion between methodology and practice.
4 2. Economics Rules
Throughout Economics Rules, Rodrik is developing the argument that economics is at its core a legitimate academic field that offers valuable understanding of social reality. Therefore, fundamental change is unnecessary. Although Rodrik criticizes economists who fail to practice economics with the right amount of rigor and modesty, he is equally critical of those who dispute the legitimacy of orthodox economics without understanding its strengths. I will give a summary of his arguments, by dividing them in four parts. First I will discuss his refutation of the critique of models, then his account of the positive features of mathematical modelling in economics, then his account of the mapping of models to reality and finally his defense of economics in the face of its critics.2
2.1 In defense of mathematical models
Rodrik starts with the topic that will be of key importance throughout the whole book. That is the topic of mathematical modelling and its role in economics. This is of such importance because mathematical modelling is a central part of the work of economists. Intellectual debate in economics largely focuses on the validity and applicability of models. The topic also shows the tension between economics and its critics. Rodrik believes that the great strengths of economics follow from its use of models. However, the use of mathematical models by economists is at the same time a major target of critique:
"For critics, economists' reliance on models captures almost everything that is wrong with the profession: the reduction of the complexities of social life to a few simplistic relationships, the willingness to make patently untrue assumptions, the obsession with mathematical rigor over realism, the frequent jump from stylized abstraction to policy conclusions."3
The points of critique raised against the practice of mathematical modelling are thus: simplisticness, lack of realism, obsession with mathematics and overeagerness to jump to policy conclusions. Rodriks aims to refute these criticisms of the way in which economists use models. He believes that, except for the last one, these points of criticism actually point out the strengths of economics, and some advantages it has over other social sciences.
We can start with the accusation of simplisticness. This is rhetorically countered by Rodrik by retelling a parable of the Argentinian author Borges. The parable is about the cartographers of an empire who want to make a map that is as realistic as possible. They end up with a scale 1:1 map, which is of little practical use. The lesson for economics is that the simplicity of models is a necessary
2 This is broadly consistent with the chapters of Economics Rules. Only the last three chapters are merged into
one.
5 feature for them to be useful. Of course social reality is very complex, but putting all this complexity in a model is probably impossible, and will otherwise just make the model as difficult to understand as the real world.
Thus, Rodrik argues, the reduction of complex social life to the simplicity of a model is necessary to give a useful analysis. The level of reduction that is needed depends on the object of analysis. A model needs to be so simple that it is capable of capturing "the most relevant aspects of
reality in a given context"4. However, it must not be too simple. If causal influences that matter to the problem at hand are left out of the model that is used, a flawed analysis results. It is therefore dangerous to mistake a simplified model for the model of social reality. If a model is thought to exhaustively describe social reality, without there being any need to consider the context of particular situations, it necessarily oversimplifies. To prevent the pitfall of universalism, economists should develop a multiplicity of models, each applicable to different contexts.
An important part of Rodriks defense of mathematical modelling, is his claim that there actually exists a multiplicity of models in economics. He points out three different types of models: a supply-demand model, a prisoners dilemma model and a multiple equilibria model. Each describe different situations and give different economic outcomes. As such, they each simplify and are not fully realistic. For Rodrik that is not a problem, since the models in themselves are not supposed to be right or wrong, and therefore do not need to be realistic.
The question of simplicity thus leads to the related question of realism. Models do not need to be completely realistic in order to teach us about reality. Rodrik describes two possible ways to think about this unrealisticness. Models can be compared to fables and to experiments.
Models, Rodrik says, can be compared to fables5, simple and short didactic stories. Although this seems to debunk the scientific character of models, the point of comparison is that both offer illumination of a certain point through the simplicity of the medium. The point that is made can be easily remembered, even if the specifics of the model or fable are forgotten. Moreover, different fables can have contrasting morals, such that in specific situations, one first has to find the most relevant fable in order to apply its moral.
On the other hand, models can be seen as experiments according to Rodrik6. Lab experiments build physical environments in order to test hypotheses. Economic models lack this
4 Rodrik, Economics Rules, 11.
5 With reference to Nancy Cartwright, "Models: Fables v. Parables," Insights (Durham Institute of Advanced
Study) 1, no. 11 (2008): 55-82.
6 Refering to Uskali Mäki, “Models Are Experiments, Experiments Are Models,” Journal of Economic
6 physical component, instead they are a mental environment that allows for testing of hypotheses. An objection to this view could be that experiments take place in a real environment and thus teach something about reality, whereas economic models are abstract and therefore do not contain information about reality. However, Rodrik says, the difference between models and experiments is one of degree, not of kind. In both cases, a result that is obtained within the model or lab environment cannot be assumed to hold outside of it. In order to apply it to outside reality, one has to test its external validity, the extent towards which the circumstances of the phenomenon under investigation match the circumstances assumed in the model or artificially created in the lab environment.
Like fables and experiments, models are not true of reality in an all-encompassing way. However, they require some level of realism in order to be informative about reality. A model is a description of a mechanism that might be at work in reality. The task of the economist who wishes to explain a certain phenomenon, is to find the model that picks out the mechanism that brings it about. A model that is capable of doing so, has to be sufficiently realistic in its critical assumptions. Models contain the conditions under which they are applicable to reality, and thus tell us what to look for. These conditions are given in what Rodrik calls critical assumptions of the model.
In a model, all kinds of assumptions are made. Some of these do not alter the conclusion that a model gives about a certain situation. These assumptions are therefore allowed to be unrealistic. However, there are other assumptions that Rodrik calls critical. He explains this as follows:
"We can say an assumption is critical if its modification in an arguably more realistic direction would produce a substantive difference in the conclusion produced by the model."7
This point brings Rodrik back to his notion of critical assumptions. He points out that he disagrees with Milton Friedman's influential view on the realism of assumptions.8 Friedman states that economists can be indifferent as to whether their assumptions are realistic, all that matters is the predictive power of their model. Rodrik says instead, that assumptions do matter. Not all of them, but for a model to be applicable to reality, its critical assumptions need to be realistic. Therefore, a model is allowed to (and in fact must) contain many abstractions from the real situation that is analyzed with it. However, any factor that significantly influences the outcome of the model must bear a resemblance to the real world.
With simplisticness and lack of realism out of the way, the remaining points of critique of economists’ use of models are overdependence on mathematics and jumping to policy conclusions.
7 Rodrik, Economics Rules, p.27.
8 As stated in Milton Friedman, “The Methodology of Positive Economics,” in Essays in Positive Economics
7 Rodrik agrees with the last point of critique. However, he does not see this as necessarily related to the use of models. His advice to economists is therefore not to stop using models, but simply to stop jumping to policy conclusions.9
That leaves the mathematical form of modelling. The frequent use of mathematics by economists comes with the risk of fetishizing mathematics. Fetishization happens when the mathematical elegance or complexity of a model gets prioritized over its usefulness and comprehensibility. Economists might do this in order to feel clever, compare their work to natural science and scare off nosy outsiders. Rodrik acknowledges this danger, but states that economists should be able to resist these temptations and only use mathematics to lend their models clarity and consistency. These are desirable features of models, and expressing models in mathematical form is a way of guaranteeing that they have them.
2.2 The strength of mathematical modelling
After having shown that the supposed downsides of economic modelling are in fact indispensable characteristics of models, Rodrik argues for the strengths of economic models. He argues that it is precisely the modelling approach to problem-solving that makes economics a science. Not a natural science, he stresses, since it does not aim at the discovery of fundamental laws. But still, a social science. Models have four characteristics that lend scientificity to economics. They clarify hypotheses, they allow for scientific progress, they imply an empirical method and they allow for knowledge generation and sharing to happen according to shared standards instead of power or hierarchy.
First of all, models clarify hypotheses. This is important says Rodrik, because when hypotheses are simply expressed in commonsensical terms, they tend not to include all assumptions that a hypothesis requires, nor all the observable results that should obtain if the hypothesis is true. Moreover, common sense relies on our intuition about causes and effects. However, Rodrik says, many mechanisms that exist in social reality, lead to counterintuitive results. It is the mathematical formalization of economic reasoning that allows us to understand counterintuitive results, whereas our non-formal reasoning would be misled by appearances.
Secondly, models allow for scientific progress. This is not the kind of progress that people tend to attribute to the natural sciences. Economics is not working on a linear development towards a perfect universal description of social reality. Instead Rodrik compares the whole of economics to a toolkit. Models are economists’ tools. When a model is rigorously formulated and proven to be applicable to some situations, it can be added to the toolkit for future use. Thus, by coming up with
8 new models, economists add tools to the toolkit of economics, thereby allowing for the explanation of an ever expanding variety of social phenomena. In this way, Economics does not progress vertically, by moving towards one universal truth, but horizontally, by expanding the possibilities of economic analysis.10
A third advantage of models is that they stimulate empirical research. Models as considered on themselves are neither true nor false, but they “contain information about the circumstances in which they’re relevant and applicable.”11 That means that economists cannot simply infer what reality is like by looking at their favorite model, but have to engage in thorough empirical research in order to find out what model applies to a given situation. The multiplicity of models therefore, make economics an empirical science.
A last advantage of models is that they allow for criticism based on the validity of formal properties of the model. This validity is not imposed by some powerful figures, but is an objectively shared standard within the profession. Therefore, anyone within the economics profession can judge and criticize the work of any other economists on the basis of formal standards, regardless of their positions in the professional hierarchy.
2.3 Using models
Rodrik has argued that models do not necessarily describe reality. They only show mechanisms that could be taking place. Models have to be carefully applied to see if they actually describe reality. It is thus important to know what the proper procedure is of applying models to reality. The right method cannot be formalized, according to Rodrik: "these methods are as much craft as they are science."12 Rodrik emphasizes this lack of formality in the metaphors he uses. He talks about navigating among models or building maps between models and the world, both of which imply skillful activity instead of formal reasoning. Through this process one hopes to arrive at a model that "highlights the dominant causal mechanism or channels at work"13 in the situation under inquiry.
In order to find the dominant mechanism, Rodrik proposes a method of diagnostics. Suppose, he says, that your car is broken. Instead of completely taking it apart and reassembling it, you will want to know which specific part is malfunctioning and fix that. You will therefore need
10 In his review of Economics Rules, Ariel Rubinstein justly notes that this is a strange way of characterizing
scientific progress. We should then also say that there is “scientific progress” in fiction writing, since growth of our libraries of fiction expands our vocabulary for making sense of the world. See Ariel Rubinstein, “Remarks on Economics Rules by Dani Rodrik,” Journal of Economic Literature, 55 no.1 (2017): 162–172.
11 Rodrik, Economics Rules, 73. 12 Id, 83.
9 some sort of procedure to find the broken part. The same logic can be applied to development economics. Unsatisfactory growth rates should not be tackled by some standardized mix of different policy measures. Instead, the main factor that slows down growth should be identified and fixed. Therefore, different kinds of policies should be tried, to see what kind of feedback results from them. This feedback can then be used to assess which factor of the economy causes the impediment to growth. Instead of implementing a plurality of different and perhaps contrasting policies, one can then specifically target the problem with appropriate policies.
The qualities of the diagnostics approach point towards more general guidelines of model selection. Rodrik states that "models generate conclusions by pairing assumptions with mechanisms of causation."14 There are thus three things about models that allow for empirical verification: critical assumptions, causal mechanisms and outcomes. Each of these has to be sufficiently similar to some part of reality, in order for the model to be instructive. Rodrik’s use of this notion of similarity is deliberately vague, in order to maintain the fact that model selection involves craft, not just the carrying out of a formalized series of acts: "While we can imagine expressing concepts such as "similarity" in formal or quantitative terms, this formalization won't be helpful in most contexts. There is an unavoidable craft element involved in rendering models useful."15
2.4 The rights and wrongs of economics
The last three chapters of Economic Rules are spent on different kinds of critique. First, critique of economic modelling by those who would like economics to focus more on grand theorizing. Then there is Rodrik’s own critique of fellow economists who neglect the plurality of economic models and are led astray by their narrow vision of economics. Finally, Rodrik discusses several criticisms against the profession of economics and argues that these are not on point.
Throughout Economics Rules, Rodrik has stressed the importance of applying economic models to particular problems, considered within their context. This context-dependent use of models may be contrasted with a search for answers to “the big, timeless questions of economics and social science.”16 That would be the domain of grand-scale economic theorizing. Historical examples of this include the works of the classical economists, the (critique of) political economy of Karl Marx and Keynes’ General Theory. Rodrik argues that these theories do not give us the grand understanding of society that they aim to give. That is because society is not a homogeneous totality, but contains all sorts of messy facts that cannot be brought under a single general theory. In order to use grand
14 Id, 99. 15 Id, 112. 16 Id, 116.
10 theories, they have to be fitted to specific parts of reality, just like regular models. Rodrik’s verdict on theories is thus that either they teach us nothing, or they are just models in disguise.
Nevertheless, economists may fail to acknowledge the multiplicity of theories and, implicitly or explicitly, believe that there is one single way of properly practicing economics. In fact, this often happens. Rodrik discusses a blogpost of Greg Mankiw, that shows a remarkable consensus among economists in favor of free-market policies and against government interventions. Mankiw has set up a list of eight propositions about the benevolence of markets and the futility and potential harm of government meddling in economic issues. They are provided without any context, but are all accepted by more than 79% of respondents as valid propositions. For Rodrik, this is an example of an immodest attitude among economists, that can cause serious damage when economists advise on policy. It is this immodesty that led to the unwarranted optimism surrounding financial markets before the 2008 financial crash. Economics is sound as long as it sticks to the modest task of trying to fit it’s models to reality. When economists believe that they have found the right way to think about reality, they make mistakes at great public cost, and should be criticized.
In fact, many people do criticize economics. There is a variety of critical points commonly raised against the field of economics by people from outside the profession or at its margins. Rodrik has already dealt with simplisticness and lack of realism in assumptions in his discussion of models. Other common objections are that economics has a too individualistic focus and that it fails to predict social phenomena. The point on individualism misses the fact that economics is capable of analyzing human agency as constrained by all sorts of factors, and is thus not necessarily individualistic in an atomistic way. The problem with prediction is due to the nature of social science. Society does not work in a deterministic way. No social science can therefore predict with complete accuracy. All that is possible is to make conditional predictions about the influence of one factor on another, while other factors are constant.
Two more things that Rodrik deals with are the claims that economics involves illegitimate value judgements and that the field lacks pluralism. On value judgements, Rodrik argues that economists are concerned with one single value: that of allocative efficiency. Again, he appeals to the modesty that economists should therefore have when it comes to policy decisions. Economists have a valuable contribution to make, when it comes to matters of efficiency. However, economists should recognize that other values then efficiency may play a role in political decision making, and that no economist can claim to have a final say in this process. The point of lacking pluralism is simply false according to Rodrik, and is a consequence of the narrow image of economics that is presented in undergraduate courses. When the whole of economics is considered, a much larger variety of modelling approaches can be shown to exist.
11 2.5 The Rules of Economics
Overall, Rodrik has argued for a certain outlook on economics. Economists should be humble when it comes to grand-scale theorizing and policy recommendations. That is because they have no grand-scale knowledge. Instead of searching for the model, economists should work on developing a multiplicity of different models. These models are then the tools in the toolkit of the economist, who must try to fit these models to social reality. If they do this in a thoughtful and rigorous way, they can be of great use for understanding social phenomena. If economists exceed their capacities however, they make costly mistakes.
3. A structured assessment
Now that this general overview of Economics Rules is given, I will continue with an assessment of the argument it makes. To do so, I will make use of two distinctions that structure Rodrik’s work. On the one hand there is the distinction between the positive and the negative part of the book, or the apologetic and the critical. Rodrik is both giving a critique of economics that he believes to be illegitimate and a defense of proper economics. This is reflected in the title of the book: The Rights
and Wrongs of the Dismal Science. Another distinction is that between abstract methodological
reasoning and examples of real-world economic practice. Rodrik switches between arguing about abstract notions such as properties of models, mathematics in economics and realism of critical assumptions, and giving examples of how economic research that is properly done gives good results. These two distinctions give us four types of arguments by Rodrik. Immodest economics goes wrong for methodological reasons, immodest economics has led to economics going wrong in the real world, proper economics is capable of good results for methodological reasons and proper economics has been successful in the real world. This structure is visualized in figure 1. I will first briefly assess Rodriks critical points, and then deal more elaborately with the arguments he raises in defense of economics.
12 Defense of proper economics Critique of bad economics
Abstract methodological reasoning
Good and useful economic science is possible through mathematical models, if these are rightly used.
Grand-scale economic theorizing is impossible and leads to error.
Real-world examples Bretton Woods, congestion pricing, conditional cash transfer, etc.
Efficient Market Hypothesis, Washington Consensus.
Figure 1: The structure of Economics Rules 3.1 Methodological Critique
Rodrik frequently stresses that economics is bound to go wrong if it mistakes a model for the model and thereby overstretches the capacities of model-based analysis of society. We may ask why this is. What is the reason for this fundamental limitation of economics? This is because of a property of social reality, namely that it is not governed by fundamental laws: “Economics is a social science, and society does not have fundamental laws- at least, not in quite the same way that nature does. Unlike a rock or a planet, humans have agency; they choose what they do. Their actions produce a near infinite variety of possibilities.”17 This is a short but clear argument for why large-scale economic theories are bound to fail. If the whole of society could be captured in a theory, all human behavior would have to be completely determined by social structures. Rodrik states, convincingly, that this is not the case.
Although there may be other ways to talk about laws in social science18, Rodrik probably has something in mind like general laws as they function in a deductive-nomological model of explanation as described by Hempel and Oppenheim19. Here, laws are thought to hold generally and of necessity. That is, a law must apply to any phenomenon that falls within its domain, without exceptions, and cannot fail to bring about a given effect. If such laws were to exist in reality, they would rigorously bind human agency to the necessities they state. That cannot be true for Rodrik.
17 Rodrik, Economics Rules, 45
18 Kincaid does so for example in Harold Kincaid, “There Are Laws in the Social Sciences,” in Contemporary
Debates in Philosophy of Science ed. Christopher Hitchcock (Malden, MA: Blackwel, 2004), 168–186. Here he works with the notion of ‘law’ as ‘something that identifies a causal force’.
19 In Carl G. Hempel, and Paul Oppenheim, "Studies in the Logic of Explanation," Philosophy of Science 15, no. 2
13 3.2 Real world Critique
Rodrik discusses two situations in which the unwarranted belief in the general applicability of a certain economic theory causes economics to lead to bad outcomes.
“When economists confuse a model for the model, two kinds of mischief may follow. First there are the errors of omission, in which a blind spot shows up in the inability to see troubles looming ahead. (...) Then there are the errors of commission, in which fixation on a particular view of the world makes economists complicit in policies whose failure might have been predicted ahead of time.”20
An example of the first is the 2008 financial crisis. This was caused by the bursting of the bubble in the U.S. mortgage market. Rodrik states that economic models were developed that economists could have used to analyze and foresee the dangers. However, the analysis of financial markets was excessively guided by the efficient market hypothesis of Eugene Fama, that states that financial markets cause asset prices to reflect all relevant information. Moreover, innovation in financial markets made economists believe that markets were capable of solving all sorts of social problems, and government regulation had become redundant. Rodrik argues that it is the narrow focus on the efficiency of financial markets that led to the financial crisis and to economists’ inability to predict it, rather than any fundamental defect of economics.
An example of the second is the so called Washington Consensus. This refers to a set of shared beliefs among policy economists in Latin America, that the main task of development policy was to create free markets and encourage competition. Rodrik explains how a set of policy measures was derived from these beliefs, that was thought to be universally applicable to any country, no matter the specific circumstances. The refusal to take into account the particular situation caused the reforms to fail. Nowadays, this sort of Washington consensus thinking in terms of universal blueprints has been replaced in development economics by tailoring policy reforms to specific environments.
3.3 Methodological defense
Things get more interesting when we take a closer look at Rodrik’s defense of economics. How is economics capable of arriving at valid results? The following train of reasoning can be derived from Economics Rules, that may serve as an abstract methodological defense of the legitimacy of economics. 1) Economics works with models. 2) Models contain a description of mechanisms. 3)
14 Mechanisms are at work in social reality. 4) Economists are capable of applying the right models to the right situations. 5) Economists gain a deeper understanding of social reality through the model than they would have had otherwise. The two key concepts in this argumentation are models and mechanisms. These are also concepts that involve much debate in philosophy of science and economic methodology. Placing Rodrik’s views within this larger debate will help to clarify, and perhaps dispute them.
3.3.1 Models
Rodrik’s methodological defense of economics largely depends on his conception of models. In the philosophical literature on scientific models, three main questions concerning models are raised21. These are the ontological question: what are models?, the epistemological question: what kind of knowledge can we derive from models?, and the semantic question: what do models refer to and how do they do this?
The ontological question is discussed by Russo22. A possible way for a model to exist is as a representation of reality, either as a set-theoretical structure referring to a fixed reality, or as a family of probability distributions referring to a probabilistic reality. This is a common way to talk about models in the natural sciences. Here, a model continuously describes certain parts of reality. This reality may behave deterministic or probabilistic, but the model represents it as good as possible. For example: the Bohr model of the atom cannot stop being a representation of an atom, due to some change in the atom. Then the model would simply be deficient.
This is clearly not the way that Rodrik thinks about models. Any model in economics is only applicable within a certain context, such that no model can be a fixed representation of a certain part of economic reality. The way Russo characterizes such thinking about models, is by thinking about models as objects. They may be fictional entities or epistemic objects. Rodrik’s account is close to the latter: models as epistemic objects. Russo states that in the latter account: “Models are tools that we build, manipulate, and use to gain the knowledge of a given phenomenon.” This is indeed the way that Rodrik characterizes models.
The following epistemic question is then what it is precisely that these models teach us. Rodrik has two answers, based on his conception of models as fables and as experiments. On the one hand, models can be compared to fables, they teach us a general qualitative morale. On the other
21 See the Stanford Encyclopedia of Philosophy lemma on models in science: Roman Frigg, and Stephan
Hartmann, "Models in Science", The Stanford Encyclopedia of Philosophy (Spring 2017 Edition), Edward N. Zalta (ed.).
22 Federica Russo, “Model-Based Reasoning in the Social Sciences,” in Springer Handbook of Model-Based
15 hand, models can be compared to experiments. In this sense, they show us mechanisms that might be at work in reality.
The last question is what models refer to, and how this is possible. This is related to the epistemic question. In order for models to teach us something, they must somehow point beyond themselves, to reality. Since economic models do not represent reality in a fixed way, they have to be applied to it by economists. We may ask however, how this is possible. If a model does not represent reality, why would it be instructive to apply it to reality? Moreover, why would it be more justified to apply it to one situation rather than another?
Rodrik’s answer can be derived from his notions of causal mechanism, critical assumptions and his reference to Uskali Mäki’s Models and the Locus of their truth23. Mäki argues, like Rodrik, that models as taken by themselves do not have a truth value. They do not describe the real world, and therefore are neither a true nor a false description of the world. What they do describe, is a causal
mechanism. That is, they show how a series of entities or activities interact such that they bring
about a certain phenomenon. A model is then correctly applied to reality, if the causal mechanism that the model describes is in fact at work in the domain of reality it is applied to.
Rodrik himself expresses this view when dealing with critical assumptions. The critical assumptions of a model are those that affect the causal mechanism. That is why any change in critical assumptions will bring about a change in the outcome of the model. By matching the critical assumptions of a model with those of the domain of reality under investigation, Rodrik believes to be able to select the model that describes the causal mechanism that is actually at work in reality.
Such reasoning is only valid under the assumption that there actually are causal mechanisms at work in the parts of reality that economists study. We may be skeptical about economic mechanisms and argue that there are no such things, and that even Rodrik’s modest search for them is in vain. Thus, we need to deal with the question: what are mechanisms and do they exist? To answer such a question would be to give an account of economic ontology.
3.3.2 Economic ontology
The philosophical subfield of ontology is concerned with study of being and tries to give an answer to the question: what is there? This question becomes more modest when it gets reduced to economic ontology, and is only related to economic issues. Then the question becomes: what is there in the economic realm? The economic realm is defined by Mäki as “those parts or aspects of the
16 universe which are set apart as constituting the subject matter of economics.”24 Economists presuppose such an economic realm. The question is what kind of entities exist in this economic realm and what properties these have. By formulating their theories in certain ways, economists presuppose some entities to exist as relevant objects of study, and others to be irrelevant or non-existent. They are thus ontologically committed to certain entities by their theories.
In his critique of grand-scale economic theorizing, Rodrik himself engages in economic ontology by stating that economics is a social science and that society does not have fundamental laws. This is because a society is made up of humans, who have agency and whose actions ‘produce a near infinite variety of possibilities.’ Therefore, laws don’t exist in the social realm. However, we can still “talk in terms of tendencies, context-specific regularities, and likely consequences.”25 These tendencies and regularities are then discussed as mechanisms by Rodrik. This raises the question however, why these mechanisms are really existing and instructive. If human agency is free and produces a near infinite variety of possibilities, why would social reality not be chaotic and impossible to capture in mathematics? Moreover, how can we learn anything from a mechanism that we have discovered, if it may cease to exist at any time, because its existence depends on a completely arbitrary organization of human action?
3.3.3 Mechanisms
In order to refute this skeptical question, some further analysis of what a mechanism is and what brings it about is necessary. Such an analysis is offered by Daniel Little. Little shares Rodriks view that “there are no ‘laws of society’ that function ontologically like laws of nature.”26 However, he aims to give a further account of what such mechanisms are. Like Rodrik, he departs from a notion of human agency as a fundamental constituent of social reality. However, for Little this is ‘structured agency’, human behavior that is always in some way bound by the structure within which it takes place. From here some regularity may be derived. It is the fact that human agency is not radically free, but takes place relative to structures that constrain it, that makes social mechanisms occur. Humans don’t just do anything but are pushed and motivated in certain directions. Little states:
24 Uskali Mäki, “Economic Ontology: what? why? how?” in The Economic Worldview, Studies in the Ontology of
Economics (Cambridge: Cambridge University Press: 2001), 4.
25 Rodrik, Economics Rules, 45.
26 Daniel Little, “Causal mechanisms in the social realm”, in Causality in the Sciences, ed. Illari, Russo, and
17 “We can directly identify the features of purposive action within given structures that make the mechanism work. Human actions and refrainings are the ‘stuff’ of social causation, and features of human agency underwrite the ‘necessity’ of social mechanisms.”27
Social causation is thus composed of human agency. Little does not cross out individual agency as something with no explanatory value, in favor of large-scale social structures that autonomously bring things about. However, human agency can only be composed into a mechanism because it is structured. The structures ‘make the mechanisms work’.
According to Little, social scientists need to analyses both the mechanisms themselves, and the things that bring them about, in a “levelled” approach to social sciences. He believes that there are “six large areas of focus for social science research”28. Among these is the question “How are individuals formed and constituted?” Where one tries to give accounts of “social development, the acquisition of preferences, worldview, and moral frameworks.” This kind of research cannot be integrated into mathematical modelling, as things such as worldview and moral frameworks, are inherently qualitative. However, in structuring the agency of individuals, they give rise to the mechanisms that economists try to capture in their models, and can therefore not be left out of consideration in economics.
Rodrik acknowledges the possibility of social structures affecting economic outcomes, but states that economists are already in the business of integrating this into their work: “an active research program in economics does exactly this, analyzing how identities, norms, and cultural practices are shaped by the interaction of individuals with each other.”29 However, there is a problem here. If economics works by applying models to mechanisms, and it is due to social structures that these mechanisms exist in the first place, then the analysis of social structures cannot simply be integrated into economic practice. The analysis of structures is prior to and different from the economic practice of mathematical modelling. Therefore, the analysis of the structure that brings about a social mechanism, is separate from the analysis that uses a model to understand a mechanism.
Such zooming in on the methodological issues concerning models and mechanisms assumes that skeptical doubts about the validity of economics need to be refuted by means of methodology. However, on Rodrik’s account, refuting a sceptic by means of abstract methodology is both impossible and unnecessary. It is impossible because the process of discovering economic
27 Id, 278.
28 Daniel Little, “Levels of the social”, in Philosophy of Anthropology and Sociology, ed. Turner, and Risjord
(Amsterdam: Elsevier, 2006), 43-372.
18 mechanisms cannot be formalized, but needs to be carried out by the skilled economist. It is also unnecessary, because economists are actually working on and succeeding in gaining a better grasp of economic reality through the use of models. When it comes to economics, the proof of the pudding is in the eating, and economics justifies itself not through an indisputable method but by showing that it can come up with results of an evident validity.
3.4 Real world defence
Rodrik does not only lay out a methodology. Economics Rules is full of examples from real-world economic practice. In his introduction Rodrik gives three examples of how economics has succeeded in doing good work. First of all there is the instalment of the Bretton Woods international monetary order, which Rodrik accredits to John Maynard Keynes and Harry Dexter White. Then there is the congestion pricing policy, which is the brainchild of William Vickery. Finally, Rodrik mentions the system of conditional cash transfer that Santiago Levy thought out as an antipoverty program. Rodrik declares each of these a success and concludes: “When economists get it right, the world gets better.”30 In this way, Rodrik starts from the point that proper economics is possible. The methodological reasoning he engages in later is not necessary for this point, but just shows how proper economics can be done.
What is remarkable about his examples however, is that these are all interventions in reality, instead of descriptions of what was already there. Rodrik describes them as follows: “in each case, economists remade part of our world by applying simple economic frameworks to public problems.”31 In this way, economists do not need to know what mechanisms exist in reality, but what economic mechanisms can be created in reality as solutions to certain problems. This is at odds with Rodrik’s method of mapping models to reality. There he is concerned with applying the right economic models to given situations that are actually going on. He even states that: “(…) the models are meant to describe what actually happens, not what should happen. There are no value judgements in this kind of analysis”32 However, when the question is how we can use economic models to create a desirable reality, economics is no longer concerned with what actually happens, but with what might possibly happen if a model were to be implemented. This makes the idea that there are no value judgements involved quite strange. Of course the question of how we want to give form to social reality involves value judgements.
30 Rodrik, Economics Rules, 5. 31 Id, 4.
19 What is telling in this context is Rodriks metaphor for growth diagnostics. He states that if a car is broken, you do not completely take it apart and put it back together, but try to find what specific part of the car’s mechanism is dysfunctional and fix that33. The comparison of an economic system with a car allows for a view on the economy as something that is not simply given, but that is constructed as an artificial machine build to serve human needs. Such a machine could then also be reassembled and rebuild in many other ways than it is at the present.
3.4.1 Models as blueprints for socio-economic machines
This way of thinking about economics in terms of the functioning or malfunctioning of a machine can be found in the work of Nancy Cartwright. Like Rodrik, she takes issue with the notion of economic laws in terms of generally applicable regularities. Unlike Rodrik, she explicitly discusses the regularities that do emerge in terms of machines. These machines are assemblies of entities with different causal capacities, that are put together in order to create some stable effect. In this sense, they are like real machines where parts are put together in such a way that the machine functions. About economics she says: “regularities are a consequence of the repeated successful operation of a
socio-economic machine.”34
What is important to note, is that if economics can be compared to a machine in this way, there is no longer only a binary opposition between a right and a wrong theory, or a correct and an incorrect application of a model. Instead of looking at one particular part of the economy, we may also ask in what way the economic machine is constructed and how it may be reassembled. Cartwright uses a car analogy too:
“This is the point of Trygve Haavelmo’s example of the relation between the height of the throttle and the speed of the auto. This is a useful relation to know if we want to make the car go faster. But if we want to build a better car, we should aim instead at understanding the capacities of the engine’s components and how they will behave in various different arrangements.” 35
This opens up a new range of possibilities for economics, one that Rodrik does not discuss. Besides analysing existing economic mechanisms, economics may contribute to the creation of economic mechanisms. Instead of fixing a broken car, where there is a single objective, making the original car work, economics may be conceived as dealing with a variety of cars that could possibly be designed.
33 Rodrik, Economics Rules, 85.
34 Nancy Cartwright, “Ceteris Paribus Laws and Socio-Economic Machines”, The Monist 78, No. 3 (1995): 278. 35 Id, 279.
20 For Cartwright, designing blueprints for socio-economic machines is about the assembly of stable causal factors. Once we have some knowledge of such causal factors, we can use economic reasoning to build models out of them. Economic analysis is a valid way of assembling causal factors into blueprints for socio-economic machines. In that sense, Cartwright is not that far removed from Rodrik’s view. Not yet existing economic mechanisms can be the subject of economic analysis, from which one can conclude that such a mechanism may be successfully implemented. However, she is more explicit than Rodrik about the normative dimension involved in using economics to intervene in social reality.
3.4.2 The performativity thesis
A more radical way to think about the implementation of not yet existing economic mechanisms in reality, is through the performativity thesis. The notion of performativity entails that an economic model or theory does not describe a given reality, but ‘performs’ its content. That is, it directs reality in some way towards the thing it describes. For example, Mackenzie argues that the Black-Scholes-Merton model of option pricing did not describe true option prices when it was formulated36. However, it became a proper model for option pricing by being accepted by option traders as a helpful tool. It was thus not it’s empirical adequacy, but it’s integrability into the reality of the trading floor, that made the Black-Scholes-Merton model a good model for option pricing. Thinking about economic models along these lines, changes the way we judge and apply them. A good model is then not one that is true, or a useful tool to understand the world, but one that is successful in becoming real.
This thought is developed by Callon37. He talks about a “pragmatic turn” that has lead proponents of performativity to characterize truth as success. In discussing Mackenzie’s work, he states: “We could say that the formula has become true, but it is preferable to say that the world it supposes has become actual.”38 Callon calls this becoming actual an agencement, about which he says: “The term agencement is a French word that has no exact English counterpart. In French its meaning is very close to “arrangement” (or “assemblage”). It conveys the idea of a combination of heterogeneous elements that have been carefully adjusted one another.”39 Of course, not every model can become real. Some models may fail to actualize themselves in reality as a stable
agencement, because of some inherent errors. However, there is definitely more than one way to
36 Donald Mackenzie, “Is Economics Performative?” in Do Economists Make Markets? On the Performativity of
Economics, ed. Mackenzie, Muniesa, and Siu, (Princeton: Princeton University Press, 2008).
37 Callon, Michel. “What Does it Mean to Say that Economics is Performative?” in Do Economists Make
Markets? On the Performativity of Economics, ed. Mackenzie, Muniesa, and Siu, (Princeton: Princeton University Press, 2008).
38 Id, 14 39 Id, 13
21 create economic reality. That insight opens up whole new paths for economics, beyond merely analyzing what is going on. Callon says: “Our work , together with the actors, is to multiply possible worlds through collective experimentations and performations.”40 Thinking about economics as performative thus means that an economist is not only concerned with the world as it is, but also with the world as it may become. That makes economics much more normative than Rodrik is willing to acknowledge.
3.4.3 Growth Diagnostics
The limits on Rodrik’s normative perspective on economics become clear when we take a closer look at his work on growth diagnostics, his own prime example of how sound economic work can be done in the real world. This is the careful ‘diagnostics’ of developing economies to find out where their constraints are, and how these can be loosened. Rodrik argues that in growth diagnostics, there is no pre-determined model that guarantees growth and development. A developing country will benefit most by analyzing what economic policy measures its specific context calls for.
That would make growth diagnostics completely compatible with Rodrik’s call for modesty of economics. Instead of departing from an over-arching vision of the economy, it lets its choice of models depend on the specific situation it is dealing with. However, some distinctions need to be made before starting the diagnostics. What makes an economy a ‘developing’ economy and what makes one a ‘developed’ economy, is a question of value and worldview, not of rigorous mathematical modelling. Moreover, in their paper on growth diagnostics, Rodrik, Hausmann and Velasco state that: “An economy that is under-performing and in need of reform is by definition one where market imperfections and distortions are rampant.”41 This statement presents a very specific image of what an economy ought to be. It must have undistorted markets, since market distortion is included in the definition of under-performance. It seems that growth diagnostics is not concerned with Cartwrights idea of ‘building a better car’, or Callons notion of ‘multiplying possible worlds’, but with fixing the economy to make it as it should be. Although it is perhaps not guided by an over-arching theory, it is guided by a narrow normative view of what the economy should be like.
4. Conclusion: Too much modesty?
Rodrik has given a thorough defense of the work done in economics. His work is important since economists should give some response to criticism from outside the profession. From the public importance of economics, it follows that the public voice should count for something and specialists should be prepared to explain and defend the work they do.
40 Id, 53
22 Rodrik has done this and provides an appealing account of economics as a field that should be practiced with rigor and modesty. The havoc that over-confident economists have helped to create makes modesty seem like a desirable feature of an economist. Moreover, the complexity of social reality indeed makes general theorizing a perilous activity, that is perhaps bound to go wrong. However, when analyzing Rodrik’s defense of good economics, the need to look beyond his modest idea of economics as careful application of mathematical models becomes apparent. Both in his methodological reasoning and in his real-world examples, such a view can be shown to be too narrow.
In his methodological reasoning, Rodrik cannot ultimately explain why models can point out mechanisms, since it is not clear why there would exist any mechanisms in the first place. In order to fix this, we need to have some idea of how agency is structured in order to bring about a mechanism. Thinking about such structure requires one to look beyond the models themselves.
Moreover, when considering Rodrik’s real-world examples of successful economics, it becomes clear that these contain many examples of models that do not describe a reality, but create one. Creating a reality involves the normative question of asking what the world should be like. The modest approach that Rodrik defends does not include this question, and thus deals with a narrow normative view of the economy.
Therefore, economists should not be as modest as Rodrik claims they should be. Neither should they naïvely engage in the grand-scale theorizing that Rodrik criticizes. Instead, they should try to look beyond the limits that Rodrik sets and ask what makes the mechanisms they study come about and what the world that economists help create should look like.
Rodrik sets out to clarify the critical discussion on economics. In a way, he also tries to end it, by showing that criticism on economics misses the point. However, it is not the case that Rodrik has once and for all cleansed economics of its shortcomings. A close scrutiny of his methodological approach, shows that his call for modesty has its problems. That means that there is work left to do for economic methodologists, who should maintain a critical discussion with those who practice economics.
Bibliography
Callon, Michel. “What Does it Mean to Say that Economics is Performative?” In Do Economists Make
Markets? On the Performativity of Economics, edited by Mackenzie, Muniesa and Siu.
Princeton: Princeton University Press, 2008.
Cartwright, Nancy. “Ceteris Paribus Laws and Socio-Economic Machines.” The Monist 78, no. 3 (1995): 276-294.
23 Cartwright, Nancy. "Models: Fables v. Parables," Insights (Durham Institute of Advanced Study) 1, no. 11
(2008): 55-82.
Friedman, Milton. “The Methodology of Positive Economics.” In Essays in Positive Economics (Chicago: University of Chicago Press, 1953).
Frigg, Roman and Stephan Hartmann. "Models in Science", The Stanford Encyclopedia of Philosophy, edited by Edward N. Zalta. 2017.
Hausmann, Ricardo, Dani Rodrik, and Andrés Velasco. “Growth Diagnostics.” (2005).
Hempel, Carl and Paul Oppenheim. "Studies in the Logic of Explanation," Philosophy of Science 15, no. 2 (1948): 135-175.
Kincaid, Harold. “There Are Laws in the Social Sciences.” In Contemporary Debates in Philosophy of
Science, edited by Christopher Hitchcock, 168–186. Malden, MA: Blackwel, 2004.
Little, Daniel. “Causal mechanisms in the social realm.” In Causality in the Sciences, edited by Illari, Russo, and Williamson, 273-295. Oxford: Oxford University Press, 2011.
Little, Daniel. “Levels of the social.” In Philosophy of Anthropology and Sociology, edited by Turner and Risjord, 343-372. Amsterdam: Elsevier, 2006.
Mackenzie, Donald. “Is Economics Performative?” In Do Economists Make Markets? On the
Performativity of Economics, edited by Mackenzie, Muniesa and Siu. Princeton: Princeton
University Press, 2008.
Mäki, Uskali. “Economic Ontology: what? why? how?” In The Economic Worldview, Studies in the
Ontology of Economics, 3-15. Cambridge: Cambridge University Press, 2001.
Mäki, Uskali. “Models and the Locus of Their Truth,” Synthese, 1, no. 180 (2011): 47-63.
Mäki, Uskali. “Models Are Experiments, Experiments Are Models,” Journal of Economic Methodology 12, no.2 (2005): 303-15.
Rodrik, Dani. Economics Rules. New York: W.W. Norton & Company, 2015.
Russo, Federica. “Based Reasoning in the Social Sciences.” In Springer Handbook of
Model-Based Science, 953–970. Cham: Springer, 2017.
