Week 4 CHAPTER 2 methodologies of positive economics
Pikkety: prejudice restricts empirical effort!
Keynes – Tinbergen debate
In the 1930’s econometrics was introduced and Tinbergen spearheaded its use. In an analysis of
business cycles, he first used econometric methods, which sparked great debate. Keynes critique was
that total completeness of a model had to be guaranteed i.e. the non-exclusion of any relevant
variable, if the model was to be used in good measurement. Therefor it was solely a model of
measurement or descriptive nature and not of discovery, criticism or prediction. He also argued that
some methods are not measurable at all and therefor useless for econometrics. There was not yet
any guarantee against invariance (that predicitons will hold over time). Furthermore, functional form
of models was still very constrained and there was no account for the problem of whether there was
invariance of future predictions. The base of Keynes arguments was that econometrics was not yet a
pure scientific method like mathematics like it claimed to be. Economic theory addressing all possible
variables was still more important. Econometrics was “statistical alchemy”.
Economic Laws
Haavelmo’s “Probability approach” to econometrics detailed how to find laws outside a laboratory.
He did so by finding a degree of persistence in relationship and invariance among economic factors.
The difficulty in finding credible economic laws, is that they are not theoretically deductible, due to
economics’ nature as a social science, but rather must be empirically observable. Because not all
factors effecting a dependent factor are known, “factual influence” can be well established but
“potential influence” is not known. Direct causality is observable but not knowing all variables in play
constrains total awareness of influences.
Measurement without theory debate (method comparison cowles committee and national bureau)
NBER had an empirical approach to science Cowles an econometric. Koopmans critique of
econometric measurement without theory was encompassed in three main arguments. The first, is a
scale argument. The complexity of economic phenomena require many variables to be in
equilibrium, without theory it would be impossible to address the right structural relationships.
Econometric techniques provided no evidence for the inclusion of certain variables as opposed to
others. Second, without structural relationships being established, the invariance of relationships
across time is not certain. Like Keynes he noted that current parameters measured can change over
time. Ironically Robert Lucas in his lucas critique criticized the Cowles commission for exactly the
same thing. Using parameters that are not structural but dependent on the model itself. In his third
argument he also sided with Keynes that econometrics could not be used for discoveries due to the
nature of the method. Moreover it could not test itself and other independent data was needed to
do this.
Milton Friedman’s methodology of positive economics (the alternative to the Cowels commission)
The Cowles committee implemented increasingly comprehensive models so to be able to incorporate
as many influencing variables in the analysis as a reaction to Keynes and Haavelmo’s critique. But,
this model called “Kleins 15 equation model” predicted at an accuracy rate less than a very simple
naïve model. It must be noted that there was critique that this very naïve model only did so because
it had very limited functional qualities. Milton Friedman disapproved of the Cowles comprehensive
model but approved of its test of validity through the naïve model, as a method of comparison. He
was of the opinion that a high correlation coefficient was not all that mattered. A good fit depends
Pikkety: prejudice restricts empirical effort!
Keynes – Tinbergen debate
In the 1930’s econometrics was introduced and Tinbergen spearheaded its use. In an analysis of
business cycles, he first used econometric methods, which sparked great debate. Keynes critique was
that total completeness of a model had to be guaranteed i.e. the non-exclusion of any relevant
variable, if the model was to be used in good measurement. Therefor it was solely a model of
measurement or descriptive nature and not of discovery, criticism or prediction. He also argued that
some methods are not measurable at all and therefor useless for econometrics. There was not yet
any guarantee against invariance (that predicitons will hold over time). Furthermore, functional form
of models was still very constrained and there was no account for the problem of whether there was
invariance of future predictions. The base of Keynes arguments was that econometrics was not yet a
pure scientific method like mathematics like it claimed to be. Economic theory addressing all possible
variables was still more important. Econometrics was “statistical alchemy”.
Economic Laws
Haavelmo’s “Probability approach” to econometrics detailed how to find laws outside a laboratory.
He did so by finding a degree of persistence in relationship and invariance among economic factors.
The difficulty in finding credible economic laws, is that they are not theoretically deductible, due to
economics’ nature as a social science, but rather must be empirically observable. Because not all
factors effecting a dependent factor are known, “factual influence” can be well established but
“potential influence” is not known. Direct causality is observable but not knowing all variables in play
constrains total awareness of influences.
Measurement without theory debate (method comparison cowles committee and national bureau)
NBER had an empirical approach to science Cowles an econometric. Koopmans critique of
econometric measurement without theory was encompassed in three main arguments. The first, is a
scale argument. The complexity of economic phenomena require many variables to be in
equilibrium, without theory it would be impossible to address the right structural relationships.
Econometric techniques provided no evidence for the inclusion of certain variables as opposed to
others. Second, without structural relationships being established, the invariance of relationships
across time is not certain. Like Keynes he noted that current parameters measured can change over
time. Ironically Robert Lucas in his lucas critique criticized the Cowles commission for exactly the
same thing. Using parameters that are not structural but dependent on the model itself. In his third
argument he also sided with Keynes that econometrics could not be used for discoveries due to the
nature of the method. Moreover it could not test itself and other independent data was needed to
do this.
Milton Friedman’s methodology of positive economics (the alternative to the Cowels commission)
The Cowles committee implemented increasingly comprehensive models so to be able to incorporate
as many influencing variables in the analysis as a reaction to Keynes and Haavelmo’s critique. But,
this model called “Kleins 15 equation model” predicted at an accuracy rate less than a very simple
naïve model. It must be noted that there was critique that this very naïve model only did so because
it had very limited functional qualities. Milton Friedman disapproved of the Cowles comprehensive
model but approved of its test of validity through the naïve model, as a method of comparison. He
was of the opinion that a high correlation coefficient was not all that mattered. A good fit depends
