Economic Methodology
Italics = lecture, normal = book
Economic Methodology investigates the nature of economics as a science. It is the philosophy of
science for economics.
Economic Methodology examines the basis and grounds for the explanations that economists give to
answer “why”-questions about the economy. (f.ex. shifting of demand and supply curve is an answer
to the question of why prices change, and economic methodology attempts to understand the
specific role these relationships play in an explanation.)
Economic Method is not the same, it tries to answer “how”-questions, and concerns techniques and
tools used by economists when making their explanations and descriptions.
Descriptive economic methodology aims to describe the different types of economic research
practices and their results. Also called positive methodology. (concerns question of how science is
actually practiced)
Prescriptive economic methodology distinguishes between good and bad explanations in economics
and considers how good explanations should be formulated. Also called normative methodology.
(concerns question of how science ought to be practiced)
Chapter 1 – The Received View of Science
The received view (also the standard view) derives from the logical positivist program in philosophy
of science.
The logical positivist program (logicism = scientific language, such as math; positivism = empiricism =
the idea that knowledge comes from experience) was a philosophical movement that lasted from the
1920s to the 1950s. They dominated thinking about philosophy of science in the first half of the 20th
century. Much of the direction in philosophy of science today is a reaction against the logical
positivists. Their main aim was to demarcate scientific knowledge, to distinguish science from pseudo-
science, and to remove any kind of metaphysical or imagined content from scientific knowledge. They
only accepted analytic or synthetic a posteriori propositions as scientific knowledge.
For the logical positivists, the only source of knowledge is experience, especially from the senses
(=empiricism).
Analytic and synthetic a posteriori propositions
Analytic sentence = a sentence in which the predicate is already given in the subject. It can be found
by analysing the subject. They are tautological – true by definition. They are explanation sentences
which don’t add any new knowledge. Example: “All bachelors are unmarried males.”
Synthetic sentence = a sentence in which a predicate is added to the subject, which is not already
given in the subject. They are extension sentences, and thus add new knowledge. These sentences
derive from experience/sense perception.
Example: “This person has brown hair.”
A priori sentence = a sentence that is built before and independently of experience. These sentences
are universal and important for science. (They don’t depend on experience, since experience only
says that something is the case, not that something is universally true)
Example: “All balls are round.”
,A posteriori sentence = a sentence that is built after and because of experience. A synthetic
statement is synthetic a posteriori if it is shown to be true by empirical research.
Example: “This ball is blue.”
In the 18th century, Immanuel Kant introduced a third important category for science: the synthetic a
priori sentences. These statements are universally true and add new knowledge, but they are neither
shown by empirical research, nor true by definition (meaning the predicate is not given in the
subject).
Example: “Everything happens for a reason.”: “Reason” is not given in the word “everything”, so it’s a
synthetic sentence. However, the sentence is universally true and thus it cannot come from
experience.
“A straight line is the shortest way between two points.”: The words “straight line” don’t contain size,
however, the sentence is universally true.
However, at the end of the 19th century/beginning of 20th century, major scientific breakthroughs
(p.10) were not in line with Kant’s third category of propositions and led to the logical positivists
denying the existence of synthetic a priori propositions in science, and asserting that all
propositions that are not true by definition should be subjected to investigation by empirical
research.
The verifiability principle states that a synthetic/non-analytic statement is meaningful if it can be
judged to be true or false by sense perception, or in other words, when it is empirically verifiable.
Because of this principle, many statements in ethics or religion must be considered meaningless in
science.
Syntactics deals with the formal relations between signs or expressions in abstraction from their
signification and interpretation.
Semantics deals with the signification and interpretation of the signs or expressions.
Summary of the aims of logical positivists:
1. To formulate precisely central philosophical notions such as a criterion of meaningfulness (the
verifiability principle) and the distinction between analytic claims (that are true by definition)
and synthetic claims (that must be testable)
2. To develop precise definitions of central scientific notions such as theory, explanation,
confirmation, etc.
Theories and evidence
The logical positivists made a distinction between the context of discovery and the context of
justification:
Context of discovery = the way in which a theory is discovered (which could be for a variety of
accidental reasons)
Context of justification = a rational reconstruction of the theory according to the tenets of logical
positivism (on the part of the discoverer and/or anyone else developing the theory) for the purpose
of its justification.
The logical positivists argued that philosophy of science should really only concern itself with the
context of justification. For them, the context of discovery was irrelevant to establishing the
scientific value of a theory. For the logical positivists, biographical and historical data about the lives
of great scientists should have no place in any serious history of the subject.
,Another fundamental distinction drawn by the logical positivists was between theories and the
evidence, facts, and data, since theories depend upon the latter for their justification:
Scientific theories = systematic collections of concepts, principles, and explanations that organize our
empirical knowledge of the world.
In the advance of scientific knowledge, theory and evidence are given different weights and they play
different roles: the main problem for philosophy of science, and also for economic methodology
based on logical positivist thinking, is to explain the relation between them.
Syntactic view = the logical positivist understanding of theory
Axiomatization (in first-order formal language) = the proper characterization of a scientific theory
according to the syntactic view.
First-order formal language:
Symbols representing variables = denoted by x, y, z, ...
Function symbols = denoted by A(⋅), B(⋅),C(⋅), ...
Predicate symbols = denoted by A, B, C, ...
¬ = “not”
∨ = “or”
∧ = “and”
→ = “if ... then”
∀ = “for all individuals”
∃ = “for some individuals”
Read p.13-14 (from “An axiomatization reduces a theory…” to “…the above definitions are taken.”
Operationalism = an extreme form of empiricism. “In general, we mean by any concept nothing more
than a set of operations; the concept is synonymous with the corresponding set of operations.”. The
operations are the ways the term is measured. The consequence of this view is that a term will have a
different meaning when it is measured in a different way.
The nature of scientific explanation
In general, an explanation is an answer to a why question. The received view of scientific explanation
is more specific: a scientific explanation should show some event or some regularity to be an instance
of a fundamental law.
Carl Hempel (1905–1997) developed this view systematically into what is called the deductive-
nomological (DN) model of explanation (or covering-law model).
In a DN explanation, a statement of what is to be explained (the explanandum) is deduced from a set
of true statements that includes at least one law (nomos). This latter set of statements is called the
explanans. The solid line represents a deductive inference.
Explanans:
Laws: L1, ..., Lm
True statements of initial conditions: c1, ..., cn
__________________________________________________________________________________
Explanandum: E
, Example:
Why did firm x raise its price?
L1 = all monopoly firms raise price when marginal cost increases
c1 = x is a monopoly firm
c2 = marginal cost has increased
____________________________________________________________________________________________________________________
E = firm x raised its price
Problems with Hempel’s DN model:
Hempel’s DN model, which focuses merely on (logical) deduction from a law, does not require
accordingly identifying causally relevant factors. Thus, according to this model, not all deductive
inferences can be considered to be explanations.
Example:
Nobody who takes birth control pills as directed gets pregnant.
George takes birth control pills as directed.
_________________________________________________________________________________________________________________
George does not get pregnant.
The birth control pills don’t explain why George did not get pregnant. It does not matter whether or
not George took birth control pills, since he is a man. In this instance taking pills is not causally
relevant for the explanation of George not getting pregnant.
The model also requires that the generalizations in the explanans be laws. But how can we be
certain that this is the case?
“law-like” statements = statements that are just like laws – though we do not know whether they are
true. So, the idea of “law-like” is the idea of a regularity. However, many regularities appear to be
accidental.
How to distinguish law-like statements from accidental regularities?
Using a syntactic definition of a law-like statement:
All law-like statements, then, have the logical form of a universal conditional:
∀x [A(x) →B(x)]
(Read as: For all x, if x is A, then x is B. The clause introduced by “if” is known as the “antecedent,”
and the clause introduced by “then” is known as the “consequent.”)
This syntactic definition, however, is still insufficient to distinguish law-like statements from
accidental regularities.
Using a number of semantic requirements:
See example p.16
Law-like statements are required to be unrestricted generalizations.
A law-like statement has no exception.
However, each time additional semantic requirements were proposed, new arguments were
developed which escaped all existing semantic requirements. And so, it seems that a law-like
statement cannot be sufficiently well defined in a semantic way.
Symmetry thesis: On the one hand, scientific explanation involves filling in what is missing above the
inference line, on the other hand, prediction involves filling in what is missing below the inference line
According to the logical positivists, without laws, then, there are no scientific explanations and no
scientific predictions, and science can only provide descriptions of individual phenomena.
There is a difficulty, perhaps impossibility, of semantically defining a law.
This is seen as presenting a greater problem for philosophers than for scientists: philosophers
have difficulties in deciding why a general regularity ought to be considered a law, scientists
are simply concerned with determining what general regularities are laws.
Italics = lecture, normal = book
Economic Methodology investigates the nature of economics as a science. It is the philosophy of
science for economics.
Economic Methodology examines the basis and grounds for the explanations that economists give to
answer “why”-questions about the economy. (f.ex. shifting of demand and supply curve is an answer
to the question of why prices change, and economic methodology attempts to understand the
specific role these relationships play in an explanation.)
Economic Method is not the same, it tries to answer “how”-questions, and concerns techniques and
tools used by economists when making their explanations and descriptions.
Descriptive economic methodology aims to describe the different types of economic research
practices and their results. Also called positive methodology. (concerns question of how science is
actually practiced)
Prescriptive economic methodology distinguishes between good and bad explanations in economics
and considers how good explanations should be formulated. Also called normative methodology.
(concerns question of how science ought to be practiced)
Chapter 1 – The Received View of Science
The received view (also the standard view) derives from the logical positivist program in philosophy
of science.
The logical positivist program (logicism = scientific language, such as math; positivism = empiricism =
the idea that knowledge comes from experience) was a philosophical movement that lasted from the
1920s to the 1950s. They dominated thinking about philosophy of science in the first half of the 20th
century. Much of the direction in philosophy of science today is a reaction against the logical
positivists. Their main aim was to demarcate scientific knowledge, to distinguish science from pseudo-
science, and to remove any kind of metaphysical or imagined content from scientific knowledge. They
only accepted analytic or synthetic a posteriori propositions as scientific knowledge.
For the logical positivists, the only source of knowledge is experience, especially from the senses
(=empiricism).
Analytic and synthetic a posteriori propositions
Analytic sentence = a sentence in which the predicate is already given in the subject. It can be found
by analysing the subject. They are tautological – true by definition. They are explanation sentences
which don’t add any new knowledge. Example: “All bachelors are unmarried males.”
Synthetic sentence = a sentence in which a predicate is added to the subject, which is not already
given in the subject. They are extension sentences, and thus add new knowledge. These sentences
derive from experience/sense perception.
Example: “This person has brown hair.”
A priori sentence = a sentence that is built before and independently of experience. These sentences
are universal and important for science. (They don’t depend on experience, since experience only
says that something is the case, not that something is universally true)
Example: “All balls are round.”
,A posteriori sentence = a sentence that is built after and because of experience. A synthetic
statement is synthetic a posteriori if it is shown to be true by empirical research.
Example: “This ball is blue.”
In the 18th century, Immanuel Kant introduced a third important category for science: the synthetic a
priori sentences. These statements are universally true and add new knowledge, but they are neither
shown by empirical research, nor true by definition (meaning the predicate is not given in the
subject).
Example: “Everything happens for a reason.”: “Reason” is not given in the word “everything”, so it’s a
synthetic sentence. However, the sentence is universally true and thus it cannot come from
experience.
“A straight line is the shortest way between two points.”: The words “straight line” don’t contain size,
however, the sentence is universally true.
However, at the end of the 19th century/beginning of 20th century, major scientific breakthroughs
(p.10) were not in line with Kant’s third category of propositions and led to the logical positivists
denying the existence of synthetic a priori propositions in science, and asserting that all
propositions that are not true by definition should be subjected to investigation by empirical
research.
The verifiability principle states that a synthetic/non-analytic statement is meaningful if it can be
judged to be true or false by sense perception, or in other words, when it is empirically verifiable.
Because of this principle, many statements in ethics or religion must be considered meaningless in
science.
Syntactics deals with the formal relations between signs or expressions in abstraction from their
signification and interpretation.
Semantics deals with the signification and interpretation of the signs or expressions.
Summary of the aims of logical positivists:
1. To formulate precisely central philosophical notions such as a criterion of meaningfulness (the
verifiability principle) and the distinction between analytic claims (that are true by definition)
and synthetic claims (that must be testable)
2. To develop precise definitions of central scientific notions such as theory, explanation,
confirmation, etc.
Theories and evidence
The logical positivists made a distinction between the context of discovery and the context of
justification:
Context of discovery = the way in which a theory is discovered (which could be for a variety of
accidental reasons)
Context of justification = a rational reconstruction of the theory according to the tenets of logical
positivism (on the part of the discoverer and/or anyone else developing the theory) for the purpose
of its justification.
The logical positivists argued that philosophy of science should really only concern itself with the
context of justification. For them, the context of discovery was irrelevant to establishing the
scientific value of a theory. For the logical positivists, biographical and historical data about the lives
of great scientists should have no place in any serious history of the subject.
,Another fundamental distinction drawn by the logical positivists was between theories and the
evidence, facts, and data, since theories depend upon the latter for their justification:
Scientific theories = systematic collections of concepts, principles, and explanations that organize our
empirical knowledge of the world.
In the advance of scientific knowledge, theory and evidence are given different weights and they play
different roles: the main problem for philosophy of science, and also for economic methodology
based on logical positivist thinking, is to explain the relation between them.
Syntactic view = the logical positivist understanding of theory
Axiomatization (in first-order formal language) = the proper characterization of a scientific theory
according to the syntactic view.
First-order formal language:
Symbols representing variables = denoted by x, y, z, ...
Function symbols = denoted by A(⋅), B(⋅),C(⋅), ...
Predicate symbols = denoted by A, B, C, ...
¬ = “not”
∨ = “or”
∧ = “and”
→ = “if ... then”
∀ = “for all individuals”
∃ = “for some individuals”
Read p.13-14 (from “An axiomatization reduces a theory…” to “…the above definitions are taken.”
Operationalism = an extreme form of empiricism. “In general, we mean by any concept nothing more
than a set of operations; the concept is synonymous with the corresponding set of operations.”. The
operations are the ways the term is measured. The consequence of this view is that a term will have a
different meaning when it is measured in a different way.
The nature of scientific explanation
In general, an explanation is an answer to a why question. The received view of scientific explanation
is more specific: a scientific explanation should show some event or some regularity to be an instance
of a fundamental law.
Carl Hempel (1905–1997) developed this view systematically into what is called the deductive-
nomological (DN) model of explanation (or covering-law model).
In a DN explanation, a statement of what is to be explained (the explanandum) is deduced from a set
of true statements that includes at least one law (nomos). This latter set of statements is called the
explanans. The solid line represents a deductive inference.
Explanans:
Laws: L1, ..., Lm
True statements of initial conditions: c1, ..., cn
__________________________________________________________________________________
Explanandum: E
, Example:
Why did firm x raise its price?
L1 = all monopoly firms raise price when marginal cost increases
c1 = x is a monopoly firm
c2 = marginal cost has increased
____________________________________________________________________________________________________________________
E = firm x raised its price
Problems with Hempel’s DN model:
Hempel’s DN model, which focuses merely on (logical) deduction from a law, does not require
accordingly identifying causally relevant factors. Thus, according to this model, not all deductive
inferences can be considered to be explanations.
Example:
Nobody who takes birth control pills as directed gets pregnant.
George takes birth control pills as directed.
_________________________________________________________________________________________________________________
George does not get pregnant.
The birth control pills don’t explain why George did not get pregnant. It does not matter whether or
not George took birth control pills, since he is a man. In this instance taking pills is not causally
relevant for the explanation of George not getting pregnant.
The model also requires that the generalizations in the explanans be laws. But how can we be
certain that this is the case?
“law-like” statements = statements that are just like laws – though we do not know whether they are
true. So, the idea of “law-like” is the idea of a regularity. However, many regularities appear to be
accidental.
How to distinguish law-like statements from accidental regularities?
Using a syntactic definition of a law-like statement:
All law-like statements, then, have the logical form of a universal conditional:
∀x [A(x) →B(x)]
(Read as: For all x, if x is A, then x is B. The clause introduced by “if” is known as the “antecedent,”
and the clause introduced by “then” is known as the “consequent.”)
This syntactic definition, however, is still insufficient to distinguish law-like statements from
accidental regularities.
Using a number of semantic requirements:
See example p.16
Law-like statements are required to be unrestricted generalizations.
A law-like statement has no exception.
However, each time additional semantic requirements were proposed, new arguments were
developed which escaped all existing semantic requirements. And so, it seems that a law-like
statement cannot be sufficiently well defined in a semantic way.
Symmetry thesis: On the one hand, scientific explanation involves filling in what is missing above the
inference line, on the other hand, prediction involves filling in what is missing below the inference line
According to the logical positivists, without laws, then, there are no scientific explanations and no
scientific predictions, and science can only provide descriptions of individual phenomena.
There is a difficulty, perhaps impossibility, of semantically defining a law.
This is seen as presenting a greater problem for philosophers than for scientists: philosophers
have difficulties in deciding why a general regularity ought to be considered a law, scientists
are simply concerned with determining what general regularities are laws.
