Milton Friedman

1. Diff. posi. and normative, related by normative's empirical research

2. Goal is to provide system of generalizations capable of prediction

3. Test is prediction

4. Cost effectiveness

5. Realism and assumptions not good judgment test

6. Theory v=gt but no because ball dropped from 30,000

8 [sic]. Simplicity usually means easier duplication and less cost

9. Since neg. statements easier to make assumptions good to facilitate indirect test, specifying conditions theory expected to be valid, economical because often more compact yet still comprehensive e.g., leaves seek to max. sunlight (ass.) vs. rules

10. Each occurrence has some features peculiarly its own and not covered by explicit rules. So crucial ass. are those relevant to judging or applying the model. Help decide which tests valid.

11. Ass. of hyp. indirect test insofar as ass. can themselves be regarded as implications of hyp. Indirect because ass. generally give rise to class of phenomena diff then class hyp. was designed for. ALso ass. kinship with other tested hyp. so relevant validity e.g. hyp. disc. closely related to degree of monopoly so if competitive disc. only to extent effects output due to harmony. This assumes single-minded pursuit of pecuniary profit which is an ass. used in several hyp. So economist might think reasonable here to. Sociologist wouldn't.

12. Marshall ortho unrealistic and dismal science. Friedman says criticism of this type is largely beside the point unless offers hyp. generating better prediction. Usually not so accompanied but just based on observation of discrepancies between ass. and real world. But no model can be realistic totally or it would be the whole. Theory is the way we perceive facts.

13. Marshall some say assumed perfect competition. Clearly this does not exist. Friedman feels Marshall did not ass. But posited one extreme. No observed demand curve will ever be horizontal so est. elasticity will always be finite so relevant? Is it sufficiently large to call infinite? No one can say. As marshall states ? must be settled by convenience of particular discussion. So Marshall's engine (model) works best for firms that can be used as if perfect comp. Need more general to cover both cases 91) where differentiation of product or fewness of number make essential diff. and (2) where it does not. This is Chamberlin and Robinson's imp. or monop. theory - unfortunately has not attributes to make a useful gen. theory, can't handle groups of firms (industries) because insists differentiation of product is essential definition of close substitute or substantial gap evade issue and make it fuzzy.

14. Yes because test old hyp. and build new. Formation of hyp. is creative act, its essence is the vision of something new in familiar material. Process is psychological.
return to books