Econometric Evaluation of Social Programs Part I: Identification - - PowerPoint PPT Presentation

econometric evaluation of social programs part i
SMART_READER_LITE
LIVE PREVIEW

Econometric Evaluation of Social Programs Part I: Identification - - PowerPoint PPT Presentation

Problems Econometric Evaluation of Social Programs Part I: Identification James J. Heckman and Edward J. Vytlacil Econ 312, Spring 2019 Heckman and Vytlacil Identification Problems Problems Identification Problems: Determining Models From


slide-1
SLIDE 1

Problems

Econometric Evaluation of Social Programs Part I: Identification

James J. Heckman and Edward J. Vytlacil Econ 312, Spring 2019

Heckman and Vytlacil Identification Problems

slide-2
SLIDE 2

Problems

Identification Problems: Determining Models From Data:

  • The identification problem asks whether theoretical constructs

have any empirical content in a hypothetical population or in real samples.

  • Specifically, consider a model space M.
  • This is the set of admissible models that are produced by some

theory for generating counterfactuals.

Heckman and Vytlacil Identification Problems

slide-3
SLIDE 3

Problems

Identification Problems: Determining Models From Data

  • Elements m ∈ M are admissible theoretical models.
  • Map g : M → T maps an element m ∈ M into an element

t ∈ T.

  • Let the class of possible information or data be I.
  • Define a map h : M → i ∈ I.

Heckman and Vytlacil Identification Problems

slide-4
SLIDE 4

Problems

Schematic of model (M), data (I), and target (T) parameter spaces

I (Data) M (Model) T (Target) h g f Are elements in T uniquely determined from elements in I ? Sometimes T = M. Usually T consists of elements derived from M.

Heckman and Vytlacil Identification Problems

slide-5
SLIDE 5

Problems

  • Let Mh(i) be the set of models consistent with i.
  • Mh(i) = h−1({i}) = {m ∈ M : h(m) = i}. The data i reject

the other model M\Mh(i).

  • By placing restrictions on models, we can sometimes reduce the

number of elements in Mh(i).

  • Going after a more limited class of objects such as features of a

model (t ∈ T) rather than the full model (m ∈ M).

  • Let Mg(t) = g −1({t}) = {m ∈ M : g(m) = t}
  • f : I → T with the property f ◦ h = g are (a) h must map M
  • nto I and (b) for all i ∈ I, there exists t ∈ T such that

Mh(i) ⊆ Mg(t).

Heckman and Vytlacil Identification Problems

slide-6
SLIDE 6

Problems

Figure 5A: identified model

I M T h g f i t

Mg (t) Mh (i)

Mh(i) = Mg (t) Heckman and Vytlacil Identification Problems

slide-7
SLIDE 7

Problems

Figure 5B: nonidentified model

I M T h g f i t

Mg (t) Mh (i)

Mh(i) ⊂ Mg (t) Heckman and Vytlacil Identification Problems