health policy (8) hbs buseinss adminstartion undergrad (2) - hks 7phds Luis Armona (taubman thurs: 10:30 - 11:45) - sylvia (review section) - oh: - fri: 4:15-5:15 (sylvia) , mon: 9-10am () check plus delineation of assumptions are made when we make particular theortical conclusions e.g. marshallian demand functions five modules (demand + consumer choice, choice under uncertainty (4), producer theory (4), equilibrium theory (7), social choices(1)) - subsidy for install solar panels in armonia - 1.equilibrium, 2.allocations of resources, 3.policy efficiency, 4.externalities - model gives testable implications which can be validated or proven wrong with data - estimate treatment effect; $y = \alpha + \beta x + \epsilon$ (local avg. treatment) - [frank wolak](https://economics.stanford.edu/people/frank-wolak): structural econometric modeling: rationales and examples from industrial organization; micro founded (consumer and producers like or not like, ) - issues of econ. casuality doesn't get resolved from putting y and x on right and left - str.model is usage of econ.theory to inform restrictions we place on data - why theory? respond to the imposition of a tax in the same way they respond to price increase (forecast the outcome of inference) ## preference relations - foundation of choices are preferences - >= binary relations (btw objects) - x=R+^2 (usually positive: short stock) - indifferent := A>=B, B>=A - strictly prefer := A>=B, not B>=A - sets in pref. space: I_y = {x \in X: x~y}: set of bundles that I like just as much as y - U_y = {x \in X: x>y}, L_y = {x \in X: y>x} - lower level set is like less or indifferent; where as indiff set is curve - 2.2kg of applies, indiff set:2.2-2.9, upper level set: x>=2 - complete if x>=y or y>=x - complete: in amazon, we can't assume preference as there are millions of products - transitive: x>=y, y>=z -> x>=z - rational: complete and transitive - choice rule: mapping $C: {B} \rightarrow S$ where S is non-empty subset of B (S may not be singleton due to indifference) - weak axiom of revealed preferences (WARP): x, y \in B, x \in C(B), then for any B' where x, y \in B' and y \in C(B'), (independence of irrevalnt preference: iia); weak cuz general - C*(B) = x in B, x>=y, - rationality is fundamental as without it we can't get WARP. WARP, C*(B) preference is not rational - preference stable to be rational?; can have choice set warp but not rational (not rational but always point to apple) - irrational QQ. gives nothing (if x is not in B') monotonic but not strongly monotonic: leontieff min(x,y) - need both of them to bigger more is better but not; I only neeed a littlbe bit of one thing to prefer that bundle; need to have everything more monotonic but not strongly montonic in eng: monotonic for every dimension (e.g. washing machine - 3개 이상은 불필) - LNS:= local non-satiated (close neighborhood):=preference are sufficiently rich (incrementally improve); 정수사과 예제는 2.3근처 동그라미 그릴수 있으므로 lns QQ. is leontieff LNS? convexity of pref: upper level set is convex (pref is cvx if upper set is ; intermediate ); not exterme convex ; strictly convex - utility functions {x|u(y) =u(y)}; monotone and convex (between curve is still in the upper level set), strictly monotone (holding fixed - strongly - unless we have vertical/horizontal line) - utility functions: dense (cvx, LNS) - next class: how to systematically relate preferences to numeric function - binary vs utility ~ partial vs general equilibrium (everything affects everything else ) - lns -> mon. -> str mon. (leontiff is lns)