THE PERSEVERANCE OF INDUCTIVE INFERENCE

Pascal Peters 3960463

The Perseverance of Inductive Inference: A battle led by Robert Sugden

Pascal Peters Utrecht University School of Economics 3960463

1

THE PERSEVERANCE OF INDUCTIVE INFERENCE Pascal Peters

3960463

Abstract

This paper will be a supportive commentary on Robert Sugden’s article ‘credible worlds: the status of theoretical models in economics’. We consider the choice of Sugden to peel away the layers of two very famous models in economics; the ‘market for lemons’ model by George Akerlof and the ‘checkerboard city for racial segregation’ mode by Thomas Schelling. We consider the role of logical empiricism and the econometric revolution on how these two models could be considered. Furthermore, we observe an additional model; ‘Efficiency wage’ model to conclude that the best approach to modelling is inductive inference. Key words: Logical empiricism, econometric revolution, inductive inference

2

THE PERSEVERANCE OF INDUCTIVE INFERENCE Pascal Peters

3960463

Introduction

In the article ‘Credible worlds: the status of theoretical models in economics’ by Robert Sugden, the author claims that he will take an theoretical economist’s approach to bridging the gap between model worlds created by economist and the real world. Although sticking to his word, it has proven inevitable to take a historical voyage though the various eras of economic methodology. In his piece, he defends the heavily discussed declaration that abstract models created by economists can and do say something about the real world. With the use of two models; George Akerlof’s ‘market for lemons’, and Thomas Schelling’s ‘checkerboard city’, he portrays his perspective and shows that modellers create counterfactual worlds. Sugden further explains that the only way of transitioning from this world to the real world is by inductive inference. After discussing the credibility of models he also concludes that our confidence in the accuracy of these models should increase the more these models account for what could have been true. (Robert Sugden, 2000) The two well-­‐known and well-­‐regarded models by Akerlof and Schelling are considered by Sugden to be conceived more by ‘conceptual exploration’ instead of ‘empirical theorizing’. Instead of focusing on the relationship between the model world and the real world the two authors place more emphasis on investigating the internal properties of the model and how they interact. With this in consideration, Sugden approaches the two models from various methodological perspectives. These consist of instrumentalism and the use of models as metaphors and caricature. For Akerlof’s ‘market for lemons’, Sudgen also briefly considers a Popperian point of view. (Robert Sugden, 2000) In this paper I will consider and discuss various aspects of Sugden’s article by highlighting the strengths and weaknesses and adding additional supportive statements

3

THE PERSEVERANCE OF INDUCTIVE INFERENCE Pascal Peters

3960463

where applicable. As much as Sugden wished to avoid writing his paper from a methodologist and philosophical perspective, it cannot confidently be stated that he succeeded. However, I should make clear that I do not object this unintended outcome, as I believe intertwining approaches from various fields is a more effective way to let a message resonate with the reader (Robert Sugden, 2000). With that said, in this paper I will also consider additional movements within the realm of economic methodology. After studying the role of logical empiricism within economic modelling, we will also take a look at how modelling is influenced by the econometric revolution brought about by John Maynard Keynes and Jan Tinbergen. To conclude, we will observe how the afore mentioned aspects play a role in the ‘efficiency wage’ model. From this it should become apparent that inductive inference is indeed the best solution.

Sugden’s Approach Sugden commences by peeling back the layers of the two famous models explored by George Akerlof and Thomas Schelling. He does this in an attempt to show the difficult transition between the model world and real world, which is vaguely and indirectly considered by both economists. He starts with Akerlof’s ‘market for lemons’ model, which is meant to show us how asymmetric information influences the functioning of various markets in the real world. This should be accomplished by showing how asymmetric information could influence a model world of the automotive market. After quoting Akerlof, ‘the automobile market is used as a finger exercise to illustrate and develop these thoughts. It should be emphasized that this market is chosen for its concreteness and ease in understanding rather than for its importance or realism’, Sugden makes a strong observation and states that Akerlof is talking about the

4

THE PERSEVERANCE OF INDUCTIVE INFERENCE Pascal Peters

3960463

transition from real world to model world. Sugden explains that an ideal representation of the real world would be to put forth an empirical case study about the automotive industry. However, this clashes with the statement Akerlof gave with regard to the importance and realism for his theory. This is due to the fact that a case study must be realistic no matter how unrepresentative it is. This leads Sugden to conclude that it may be the real market that is unimportant but that it is the model that is unrealistic. It could be said that Sugden is suggesting that a lot of realistic explanatory value is taken away in Akerlof’s model. This redirecting of the readers attention is a strong opening for Sudgen to lay forth his claim that the best solution is an inductive inference approach. (Robert Sugden, 2000) Sugden continues by applying a similar deconstructive approach to Thomas Schelling’s ‘checkerboard model of racial sorting’. It is evident that Schelling wants to explain the unintended consequences being brought about by uncoordinated individual actions. He is trying to indicate that even though people might have non-­‐strict preferences about what race their neighbors are, the fact that everyone has a different criteria leads to a shuffling of the population. Once the dust settles it is evident and unavoidable that this has lead to the unintended full segregation of the population based on race. Sugden points out the similarity that this model has with Akerlof’s model for lemons. He states that Shelling’s model also shows that, ‘some regularity R (persistent racial segregation with moving geographical boundaries) can be found in economic or social phenomena. It also claims that R can be explained by some set of causal factors F (a common preference not to be heavily outnumbered by neighbors not of one’s own type)’. However, with all said and done this only manages to accomplish giving an ambiguous demonstration of what could also happen in real life. Simply because R occurs, F occurs and often this is simultaneously does not indicate causation. However, it

5

THE PERSEVERANCE OF INDUCTIVE INFERENCE Pascal Peters

3960463

seems that by create F into the specification of the model, Schelling manages to avoid the notion of correlation and indirectly allows for the causation. It is very counter-­‐intuitive but the author attempt to create credibility by the fact that R is represented in the model world as an extreme (complete segregation). (Robert Sugden, 2000) In order to discover what makes a model credible, Sugden first touches upon the notion of conceptual exploration. Both Akerlof’s and Schelling’s models can be seen as a fixed and protected model world which is only concerned with the internal properties of the model instead of the relationship with the real world. The value of this is that it allows for theorizing, which is practical for many reasons. From a Popperian point of view however, it does not allow for many possibilities with which the model can be falsified. This is turn does not does not allow the model or better said ‘theory’ to be a strong one. If we consider both models to be conceptual explorations we come across what Sugden considers to be counter-­‐factual arguments. Instead of directly drawing inferences from the real world, both economists consider some possible links, using careful terminology such as ‘may explain’. It is evident that neither Akerlof nor Schelling is willing to claim that their respective models are complete theories. Rather than this, their models create a premise for other economists or readers to interpret their own theory.

Logical Empiricism vs. Instrumentalism Also referred to as logical positivism, logical empiricism was a philosophical movement considered by much of the western world starting around the 1920s. It was

6

THE PERSEVERANCE OF INDUCTIVE INFERENCE

mainly concerned with exactly what the name suggests; logicism and

Pascal Peters 3960463

positism/empiricism. The former refers to the concept that logic is capable putting forward definitions for primitive mathematical concepts. The latter suggests that concrete knowledge can only be achieved from learning out of experience. Instrumentalism on the other hand follows the belief that a theory should be judged based on their predictive power within the field of its origin. When considering the two models brought forward by Akerlof and Schelling we encounter a problem. As mentioned before, it is fair enough that from experience a modeler can claim that if F occurs and R occurs in correlation, you have found a possible theory. However, the question is what has given the modelers the belief that F is a credible representation of an actual occurs in the real world. In other words, why is F considered to have the power to explain R. The first question that needs to be asked is how can we test why the suggested F variables are thought of the way that they are and whether this is fair. This is clearly impossible to do using the logical mathematics approach of logical positivism. From experience, I would claim that there is a strong correlation between rainy whether and my shoes getting wet. This does not give me the support to claim that one is caused by the other. It could be that I haven’t gone outside all day but that I am a clumsily drinker and spill my beverage almost everyday (of course pouring down to my shoes). It is still fair to claim that there is a correlation between the rain and the wet shoes but it does not consider that my clumsiness should be the actual variable considered. (Banach, 2006) In his paper, Sugden explains, ‘An alternative form of instrumentalism, perhaps more appropriate for economics, accepts that the assumptions of a theory refer to things in the real world, but maintains that it does not matter whether those assumptions are true or false.’ This is exactly what I was referring to in the previous paragraph. This is

7

THE PERSEVERANCE OF INDUCTIVE INFERENCE Pascal Peters

3960463

crucial since the assumption of a theory function as a prediction of the outcomes. Unlike a logical positivist’s, deductive mathematical approach, a change in variables in a model that should explain the real world has a different effect on the model than mathematical variables. Using the previous example, over some time I could have learned how to not be so clumsy. However, considering the mathematical variable of height as determinant for pricing of a building, it would make no sense to claim that the representation of height becomes incredible. A decrease on a rule by 1 cm today will have the same representation value today as it while tomorrow and so on (unless we change numerical calibration). (Robert Sugden, 2000)

The econometric revolution Furthermore, Sugden considers the work of Deirdre McCloskey, who focuses more on the explanatory function of models instead of its predictor power. By giving a story to a relationship (in the form of a metaphor) an undeniable causal effect has been put forth. Allan Gibbard and Hal Varian study a similar possibility, namely the use of caricatures. I believe that this concept has its strengths and weaknesses. It could be said that the exaggerated isolated representations of reality makes it easy to communicate the representation at hand. The only question is whether this is the right representation. The struggle to estimate which variables to consider in isolation is made even more difficult when one has to consider multiple (potentially hundreds) other variables also having an influential role. Do the other variables also influence the dependent variable or do they have an influence on the other independent variables, which in turn influence the dependent variable? What is the best approach to consider the influences of a variable in isolation? (Robert Sugden, 2000)

8

THE PERSEVERANCE OF INDUCTIVE INFERENCE

Pascal Peters 3960463

Jan Tinbergen pioneered the answers to these questions and John Maynard Keynes challenged his findings. It was an era that brought about the econometric revolution. However, the initial aim of Tinbergen was to find a way to portray the development of real business cycles. The use of economic data to make models representing business cycles was well favored for its strength in observing a particular variable’s influence which holding other variables constant, ceteris paribus. I will attempt to explain why an econometrics approach to Akerlof and Schelling’s models cannot be considered as representative. From the two models we understand the following. (Keynes, 1939)

Occurrence of F

Suggested influence

Occurrence of R

This suggests the ceteris paribus situation that one can observe using econometrics. However, what the models should actually be considering as a possibility is the following. (all arrows suggest possible influences that need to be considered)

Occurrence of F1

Occurrence of F5 Occurrence of F2

Occurrence of R Occurrence of F6

Occurrence of F3

9

THE PERSEVERANCE OF INDUCTIVE INFERENCE Pascal Peters

3960463

I understand that this model world (represented by the blue box) seems very exaggerated and unnecessary, but the truth is that the interactions of variables in the real world are even more numerous and complex. Simply because certain events occur in the same environment does not give justification for someone to claim to F causes R. I wish to explain my argument with an example.

Efficiency Wage Theory ‘model’ The hypothesis of this model is that in some markets the development of wages leads them to not be market-­‐clearing. It suggests that paying employees a wage that is higher than the marginal revenue product leads to workers being more productive. A higher wage leads to an employee also potential losing more if they get fired. This in provides extra incentive for them to be efficient. The term ‘efficiency wages’ was introduced by Alfred Marshall. Yet, the most influential models (concerning shirking), were brought about by Carl Shapiro and Joseph E. Stiglitz. Considering the approach of Akerlof and Schelling, we would simply say that and increase in productivity seems to be occurring at the same time as an increase in wages. From this they would conclude that increasing wages would increase productivity. This is a dangerous claim to make considering all the possible variables that influence productivity. Take job security as an example. Getting an increase in your wage from $40,000 a year to $60,000 a year would fit well within the simplistic models of Akerlof and Schelling. However, the variable of increased wages will not be as explanatory anymore when someone gets and increase in wage from $150,000 a year to $170,000. At some point the variable job security obtains a greater level of importance. This shows the interaction between independent variables ‘wage’ and ‘job security’. Imagine having to consider the

10

THE PERSEVERANCE OF INDUCTIVE INFERENCE Pascal Peters

3960463

interaction between hundreds of variables. This cannot be done efficiently when observing ceteris paribus conditions.

Conclusion In the article by Sugden, he gives a strong explanation of three different possibilities of how Akerlof and Schelling’s modeling could be understood. These are namely explanation, prediction and abduction. Each method requires a leap from the real world to the model world to explain the relationship. An inductive approach needs to be taken to allow the specific propositions in model to become more general. If we know that there is a general link between F and R, when we see them together in the real world we can expect that F will in some way explain the behavior of R. Likewise, one could also expect to find R in the real world when we see F and vice-­‐versa. This allows the relationship to become more justified but it must be kept in mind that it allows for a decrease in accuracy. This could however also be seen as an ad hoc stratagem to protect the model from potential unjustifiable statements.

Sugden takes a critical but reasonable approach to understanding the various aspects of both Akerlof and Schelling’s models. From this the reader can understand that the two economics created a model world, which has the ability to give readers their own choice of how to interpret the arguments. When considering what is right or wrong, it is important to consider who is answering the question. In the end, if many readers take the same information provided by the models and constructs a similar outcome it can be concluded that the universally excepted outcome is the right one. As economics keeps striving to define itself as a real science new things will be learn and understood. However

11

THE PERSEVERANCE OF INDUCTIVE INFERENCE Pascal Peters

3960463

inaccurate it may be, we can only make statements how we know best using what we understand best. As this changes, our minds should always be kept open for the possibility of different explanations.

Exchange Partner: Laurens Rijnja Word count: 2,611

12

THE PERSEVERANCE OF INDUCTIVE INFERENCE

References

Pascal Peters 3960463

Banach, D. (2006). The Logical Empiricist Interpretation of Scientific Method. The Logical Empiricist Interpretation of Scientific Method. Retrieved June 19, 2014, from http://www.anselm.edu/homepage/dbanach/logemp.htm Keynes, J. M. League of Nations: Professor Tinbergen's Method. Economic Journal, 47, 558-­‐577. Retrieved June 7, 2014, from http://links.jstor.org/sici?sici=0013-­‐
0133%28193909%2949%3A195%3C558%3AOP%3E2.0.CO%3B2-­‐%23

Robert Sugden (2000) Credible worlds: the status of theoretical models in economics, Journal of Economic Methodology, 7:1, 1-­‐31 http://dx.doi.org/10.1080/135017800362220

13

THE PERSEVERANCE OF INDUCTIVE INFERENCE Pascal Peters

3960463

Essay exchange Paper by Pascal Peters Reviewed by Laurens Rijnja

Style The terminology used in this paper is rather complex, but due to the clear explanation of the applicable terms, this makes for a great read. In addition, there is no terminology added, except for the explanation and commentary of previously used terminology. The models in the blue boxes give a clear illustration of what is meant, but the explanation is not as concrete as it could be in my opinion.

Structure This paper is well structured. The introduction give a clear overview of what the paper will be discussing and the conclusion gives a clear overview of the initial ‘problem’ and the material covered in between. Although the sentences connect very well during the paper, the use of indentations could improve the structural overview of the paper, otherwise the paragraphs are too long to come to a clear understanding.

Argumentation This paper leaves not much to say about the argumentation. The arguments used in the paper by Sugden are well explained and the argumentation to support his opinion are elaborated on as well. The number of own arguments however, though clear, are limited. This might be something to improve upon. Improvements In order to improve your paper, there are some things to be done. I would suggest to reread the paper again to exclude minor mistakes in spelling and grammar. Some explanations of earlier models can be simplified of elaborated on in order to give a better representation of these models. If desirable, the use of indentations may provide an improved structure. In addition, there are several references missing in the body of the paper.

Good luck on improving your paper!

14

THE PERSEVERANCE OF INDUCTIVE INFERENCE

First Draft

Pascal Peters 3960463

Introduction

In the article ‘Credible worlds: the status of theoretical models in economics’ by Robert
Sugden, the author claims that he will take an theoretical economist’s approach to bridging the gap between model worlds created by economist and the real world. Although sticking to his word, it has proven inevitable to take a historical voyage though the various eras of economic methodology. In his piece, he defends the heavily discussed declaration that abstract models created by economists can and do say something about the real world. With the use of two models; George Akerlof’s ‘market for lemons’, and Thomas Schelling’s
‘checkerboard city’, he portrays his perspective and shows that modellers create counterfactual worlds. Sugden further explains that the only way of transitioning from this world to the real world is by inductive inference. After discussing the credibility of models he also concludes that our confidence in the accuracy of these models should increase the more these models account for what could have been true.
The two well-known and well-regarded models by Akerlof and Schelling are considered by
Sugden to be conceived more by ‘conceptual exploration’ instead of ‘empirical theorizing’.
Instead of focusing on the relationship between the model world and the real world the two authors place more emphasis on investigating the internal properties of the model and how they interact. With this in consideration, Sugden approaches the two models from various methodological perspectives. These consist of instrumentalism and the use of models as metaphors and caricature. For Akerlof’s ‘market for lemons’, Sudgen also briefly considers a
Popperian point of view.
In this paper I will consider and discuss various aspects of Sugden’s article by highlighting the strengths and weaknesses and adding additional supportive statements where applicable.
As much as Sugden wished to avoid writing his paper from a methodologist and philosophical perspective, it cannot confidently be stated that he succeeded. However, I should make clear that I do not object this unintended outcome, as I believe intertwining approaches from various fields is a more effective way to let a message resonate with the reader. With that said, in this paper I will also consider additional movements within the realm of economic methodology. After studying the role of logical empiricism within economic modelling, we will also take a look at how modelling is influenced by the econometric revolution brought about by John Maynard Keynes and Jan Tinbergen. To conclude, we will observe how the afore mentioned aspects play a role in the ‘efficiency wage’ model. From this it should become apparent that inductive inference is indeed the best solution.
Sugden’s Approach
Sugden commences by peeling back the layers of the two famous models explored by George
Akerlof and Thomas Schelling. He does this in an attempt to show the difficult transition between the model world and real world, which is vaguely and indirectly considered by both economists. He starts with Akerlof’s ‘market for lemons’ model, which is meant to show us how asymmetric information influences the functioning of various markets in the real world.
This should be accomplished by showing how asymmetric information could influence a model world of the automotive market. After quoting Akerlof, ‘the automobile market is

used as a finger exercise to illustrate and develop these thoughts. It should be emphasized that this market is chosen for its concreteness and ease in understanding

15

THE PERSEVERANCE OF INDUCTIVE INFERENCE

Pascal Peters 3960463

rather than for its importance or realism’, Sugden makes a strong observation and states that Akerlof is talking about the transition from real world to model world.
Sugden explains that an ideal representation of the real world would be to put forth an empirical case study about the automotive industry. However, this clashes with the statement Akerlof gave with regard to the importance and realism for his theory. This is due to the fact that a case study must be realistic no matter how unrepresentative it is. This leads Sugden to conclude that it may be the real market that is unimportant but that it is the model that is unrealistic. It could be said that Sugden is suggesting that a lot of realistic explanatory value is taken away in Akerlof’s model. This redirecting of the readers attention is a strong opening for Sudgen to lay forth his claim that the best solution is an inductive inference approach.
Sugden continues by applying a similar deconstructive approach to Thomas
Schelling’s ‘checkerboard model of racial sorting’. It is evident that Schelling wants to explain the unintended consequences being brought about by uncoordinated individual actions. He is trying to indicate that even though people might have non-strict preferences about what race their neighbors are, the fact that everyone has a different criteria leads to a shuffling of the population. Once the dust settles it is evident and unavoidable that this has lead to the unintended full segregation of the population based on race. Sugden points out the similarity that this model has with Akerlof’s model for lemons. He states that Shelling’s model also shows that, ‘some regularity R
(persistent racial segregation with moving geographical boundaries) can be found in economic or social phenomena. It also claims that R can be explained by some set of causal factors F (a common preference not to be heavily outnumbered by neighbors not of one’s own type)’. However, with all said and done this only manages to accomplish giving an ambiguous demonstration of what could also happen in real life.
Simply because R occurs, F occurs and often this is simultaneously does not indicate causation. However, it seems that by create F into the specification of the model,
Schelling manages to avoid the notion of correlation and indirectly allows for the causation. It is very counter-intuitive but the author attempt to create credibility by the fact that R is represented in the model world as an extreme (complete segregation).
In order to discover what makes a model credible, Sugden first touches upon the notion of conceptual exploration. Both Akerlof’s and Schelling’s models can be seen as a fixed and protected model world which is only concerned with the internal properties of the model instead of the relationship with the real world. The value of this is that it allows for theorizing, which is practical for many reasons. From a
Popperian point of view however, it does not allow for many possibilities with which the model can be falsified. This is turn does not does not allow the model or better said
‘theory’ to be a strong one. If we consider both models to be conceptual explorations we come across what Sugden considers to be counter-factual arguments. Instead of directly drawing inferences from the real world, both economists consider some possible links, using careful terminology such as ‘may explain’. It is evident that neither Akerlof nor Schelling are willing to claim that their respective models are complete theories. Rather than this, their models create a premise for other economists or readers to interpret their own theory.

16

THE PERSEVERANCE OF INDUCTIVE INFERENCE

Pascal Peters 3960463

Logical Empiricism vs. Instrumentalism
Also referred to as logical positivism, logical empiricism was a philosophical movement considered by much of the western world starting around the 1920s. It was mainly concerned with exactly what the name suggests; logicism and positism/empiricism. The former refers to the concept that logic is capable putting forward definitions for primitive mathematical concepts. The latter suggests that concrete knowledge can only be achieved from learning out of experience.
Instrumentalism on the other hand follows the belief that a theory should be judged based on their predictive power within the field of its origin. When considering the two models brought forward by Akerlof and Schelling we encounter a problem. As mentioned before, it is fair enough that from experience a modeler can claim that if F occurs and R occurs in correlation, you have found a possible theory. However, the question is what has given the modelers the belief that F is a credible representation of an actual occurs in the real world. In other words, why is F considered to have the power to explain R. The first question that needs to be asked is how can we test why the suggested F variables are thought of the way that they are and whether this is fair.
This is clearly impossible to do using the logical mathematics approach of logical positivism. From experience, I would claim that there is a strong correlation between rainy whether and my shoes getting wet. This does not give me the support to claim that the one is caused by the other. It could be that I haven’t gone outside all day but that I am a clumsily drinker and spill my beverage almost everyday (of course pouring down to my shoes). It is still fair to claim that there is a correlation between the rain and the wet shoes but it does not consider that my clumsiness should be the actual variable considered.
In his paper, Sugden explains, ‘An alternative form of instrumentalism, perhaps more appropriate for economics, accepts that the assumptions of a theory refer to things in the real world, but maintains that it does not matter whether those assumptions are true or false.’ This is exactly what I was referring to in the previous paragraph. This is crucial since the assumption of a theory function as a prediction of the outcomes.
Unlike a logical positivist’s, deductive mathematical approach, a change in variables in a model that should explain the real world has a different effect on the model than mathematical variables. Using the previous example, over some time I could have learned how to not be so clumsy. However, considering the mathematical variable of height as determinant for pricing of a building, it would make no sense to claim that the representation of height becomes incredible. A decrease on a rule by 1 cm today will have the same representation value today as it while tomorrow and so on (unless we change numerical calibration).
The econometric revolution
Furthermore, Sugden considers the work of Deirdre McCloskey, who focuses more on the

explanatory function of models instead of its predictor power. By giving a story to a relationship (in the form of a metaphor) an undeniable causal effect has been put forth.

17

THE PERSEVERANCE OF INDUCTIVE INFERENCE

Pascal Peters 3960463

Allan Gibbard and Hal Varian study a similar possibility, namely the use of caricatures. I believe that this concept has its strengths and weaknesses. It could be said that the exaggerated isolated representations of reality makes it easy to communicate the representation at hand. The only question is whether this is the right representation. The struggle to estimate which variables to consider in isolation is made even more difficult when one has to consider multiple (potentially hundreds) other variables also having an influential role. Do the other variables also influence the dependent variable or do they have an influence on the other independent variables, which in turn influence the dependent variable? What is the best approach to consider the influences of a variable in isolation?
Jan Tinbergen pioneered the answers to these questions and John Maynard Keynes challenged his findings. It was an era that brought about the econometric revolution.
However, the initial aim of Tinbergen was to find a way to portray the development of real business cycles. The use of economic data to make models representing business cycles was well favored for its strength in observing a particular variable’s influence which holding other variables constant, ceteris paribus. I will attempt to explain why an econometrics approach to Akerlof and Schelling’s models cannot be considered as representative. From the two models we understand the following.

Occurrence of F

Suggested influence

Occurrence of R

This suggests the ceteris paribus situation that one can observe using econometrics. However, what the models should actually be considering as a possibility is the following. (all arrows suggest possible influences that need to be considered)

Occurrence of F1

Occurrence of F5

Occurrence of F2 Occurrence of R

Occurrence of F6 Occurrence of F3

I understand that this model world (represented by the blue box) seems very exaggerated and unnecessary, but the truth is that the interactions of variables in the real world are even more numerous and complex. Simply because certain events occur in the same environment does not give justification for someone to claim to F causes R. I wish to explain my argument with an example.

Efficiency Wage Theory ‘model’

18

THE PERSEVERANCE OF INDUCTIVE INFERENCE Pascal Peters

3960463

The hypothesis of this model is that in some markets the development of wages leads them to not be market-­‐clearing. It suggests that paying employees a wage that is higher than the marginal revenue product leads to workers being more productive. A higher wage leads to an employee also potential losing more if they get fired. This in provides extra incentive for them to be efficient. The term ‘efficiency wages’ was introduced by Alfred Marshall. Yet, the most influential models (concerning shirking), were brought about by Carl Shapiro and Joseph E. Stiglitz. Considering the approach of Akerlof and Schelling, we would simply say that and increase in productivity seems to be occurring at the same time as an increase in wages. From this they would conclude that increasing wages would increase productivity. This is a dangerous claim to make considering all the possible variables that influence productivity. Take job security as an example. Getting an increase in your wage from $40,000 a year to $60,000 a year would fit well within the simplistic models of Akerlof and Schelling. However, the variable of increased wages will not be as explanatory anymore when someone gets and increase in wage from $150,000 a year to $170,000. At some point the variable job security obtains a greater level of importance. This shows the interaction between independent variables ‘wage’ and ‘job security’. Imagine having to consider the interaction between hundreds of variables. This cannot be done efficiently when observing ceteris paribus conditions.

Conclusion In the article by Sugden, he gives a strong explanation of three different possibilities of how Akerlof and Schelling’s modeling could be understood. These are namely explanation, prediction and abduction. Each method requires a leap from the real world to the model world to explain the relationship. An inductive approach needs to be taken to allow the specific propositions in model to become more general. If we know that there is a general link between F and R, when we see them together in the real world we can expect that F will in some way explain the behavior of R. Likewise, one could also expect to find R in the real world when we see F and vice-­‐versa. This allows the relationship to become more justified but it must be kept in mind that it allows for a decrease in accuracy. This could however also be seen as an ad hoc stratagem to protect the model from potential unjustifiable statements.

Sugden takes a critical but reasonable approach to understanding the various aspects of both Akerlof and Schelling’s models. From this the reader can understand that the two economics created a model world, which has the ability to give readers their own choice of how to interpret the arguments. When considering what is right or wrong, it is important to consider who is answering the question. In the end, if many readers take the same information provided by the models and constructs a similar outcome it can be concluded that the universally excepted outcome is the right one. As economics keeps striving to define itself as a real science new things will be learn and understood. However inaccurate it may be, we can only make statements how we know best using what we understand best. As this changes, our minds should always be kept open for the possibility of different explanations.

19