Chapter 71 Econometric Evaluation of Social Programs, Part II: Using the Marginal Treatment Effect to Organize Alternative Econometric Estimators to Evaluate Social Programs, and to Forecast their Effects in New Environments ⁎
Citations
3,175 citations
3,154 citations
1,636 citations
Cites background from "Chapter 71 Econometric Evaluation o..."
...52See Heckman and Robb (1985), Heckman and Vytlacil (2007) and Matzkin (2007) for a discussion of replacement functions....
[...]
...See Herrnstein and Murray (1994), Murnane, Willett, and Levy (1995), and Cawley, Heckman, and Vytlacil (2001). (2)See Heckman, Stixrud, and Urzua (2006), Borghans, Duckworth, Heckman, and ter Weel (2008) and the references they cite. See also the special issue of the Journal of Human Resources 43 (4), Fall 2008 on noncognitive skills. (3)See Cunha, Heckman, Lochner, and Masterov (2006) and Cunha and Heckman (2007, 2009). (4)This evidence is summarized in Knudsen, Heckman, Cameron, and Shonkoff (2006) and Heckman (2008). (5)See Shumway and Stoffer (1982) and Watson and Engle (1983) for early discussions of such models....
[...]
...See Herrnstein and Murray (1994), Murnane, Willett, and Levy (1995), and Cawley, Heckman, and Vytlacil (2001)....
[...]
...See Herrnstein and Murray (1994), Murnane, Willett, and Levy (1995), and Cawley, Heckman, and Vytlacil (2001). (2)See Heckman, Stixrud, and Urzua (2006), Borghans, Duckworth, Heckman, and ter Weel (2008) and the references they cite. See also the special issue of the Journal of Human Resources 43 (4), Fall 2008 on noncognitive skills. (3)See Cunha, Heckman, Lochner, and Masterov (2006) and Cunha and Heckman (2007, 2009)....
[...]
...See Herrnstein and Murray (1994), Murnane, Willett, and Levy (1995), and Cawley, Heckman, and Vytlacil (2001). (2)See Heckman, Stixrud, and Urzua (2006), Borghans, Duckworth, Heckman, and ter Weel (2008) and the references they cite....
[...]
1,585 citations
1,206 citations
References
23,744 citations
16,198 citations
"Chapter 71 Econometric Evaluation o..." refers background or methods in this paper
...See Todd (1999, 2007, 2008) for software and extensive discussion of the mechanics of matching....
[...]
...To see the consequences of this violation in a regression setting, use Y = Y0 + D(Y1 − Y0) and take conditional 3 See the discussion in Section 8....
[...]
...See Heckman (1992) for a discussion of randomization bias in economics....
[...]
...More recent work analyzes distributions of outcomes [e.g., Aakvik, Heckman and Vytlacil (2005), Carneiro, Hansen and Heckman (2003)]....
[...]
...See Powell (1994) for a survey....
[...]
11,977 citations
"Chapter 71 Econometric Evaluation o..." refers background in this paper
...They are manifestations of a more general problem termed “Hawthorne effects” that arise from observing any population [see Campbell and Stanley (1963), Cook and Campbell (1979)]....
[...]
10,916 citations
"Chapter 71 Econometric Evaluation o..." refers methods or result in this paper
...38 In an application to wage equations, Card (1999, 2001) interprets the LATE estimator as identifying returns to marginal persons. Heckman (1996) notes that the actual margin of choice selected by the IV estimator is not identified by the instrument. It is unclear as to which segment of the population the return estimated by LATE applies. If the analyst is interested in knowing the average response ( β̄ ) , the effect of the policy on the outcomes of countries that adopt it (E(β | D = 1)) or the effect of the policy if a particular country adopts it, there is no guarantee that the IV estimator comes any closer to the desired target than the OLS estimator and indeed it may be more biased than OLS. Because different instruments define different parameters, having a wealth of different strong instruments does not improve the precision of the estimate of any particular parameter. This is in stark contrast with the traditional model with β ⊥ D. In that case, all valid instruments identify β̄. The Durbin (1954) – Wu (1973) – Hausman (1978) test for the validity of extra instruments applies to the traditional model. In the more general case with essential heterogeneity, because different instruments estimate different parameters, no clear inference emerges from such specification tests. When there are more than two distinct values of Z, Imbens and Angrist draw on the analysis of Yitzhaki (1989), which was refined in Yitzhaki (1996) and Yitzhaki and Schechtman (2004), to produce a weighted average of pairwise LATE parameters where the scalars Z are ordered to define the LATE parameter. In this case, IV is a weighted average of LATE parameters with nonnegative weights.39 Imbens and Angrist generalize this result to the case of vector Z assuming that instruments are monotonic functions of the probability of selection. Heckman and Vytlacil (1999, 2001b, 2005), Heckman, Urzua and Vytlacil (2006) and Carneiro, Heckman and Vytlacil (2006) generalize the analysis of Imbens and Angrist (1994) in several ways and we report their results in this chapter....
[...]
...38 In an application to wage equations, Card (1999, 2001) interprets the LATE estimator as identifying returns to marginal persons. Heckman (1996) notes that the actual margin of choice selected by the IV estimator is not identified by the instrument....
[...]
...38 In an application to wage equations, Card (1999, 2001) interprets the LATE estimator as identifying returns to marginal persons. Heckman (1996) notes that the actual margin of choice selected by the IV estimator is not identified by the instrument. It is unclear as to which segment of the population the return estimated by LATE applies. If the analyst is interested in knowing the average response ( β̄ ) , the effect of the policy on the outcomes of countries that adopt it (E(β | D = 1)) or the effect of the policy if a particular country adopts it, there is no guarantee that the IV estimator comes any closer to the desired target than the OLS estimator and indeed it may be more biased than OLS. Because different instruments define different parameters, having a wealth of different strong instruments does not improve the precision of the estimate of any particular parameter. This is in stark contrast with the traditional model with β ⊥ D. In that case, all valid instruments identify β̄. The Durbin (1954) – Wu (1973) – Hausman (1978) test for the validity of extra instruments applies to the traditional model. In the more general case with essential heterogeneity, because different instruments estimate different parameters, no clear inference emerges from such specification tests. When there are more than two distinct values of Z, Imbens and Angrist draw on the analysis of Yitzhaki (1989), which was refined in Yitzhaki (1996) and Yitzhaki and Schechtman (2004), to produce a weighted average of pairwise LATE parameters where the scalars Z are ordered to define the LATE parameter....
[...]
...38 In an application to wage equations, Card (1999, 2001) interprets the LATE estimator as identifying returns to marginal persons. Heckman (1996) notes that the actual margin of choice selected by the IV estimator is not identified by the instrument. It is unclear as to which segment of the population the return estimated by LATE applies. If the analyst is interested in knowing the average response ( β̄ ) , the effect of the policy on the outcomes of countries that adopt it (E(β | D = 1)) or the effect of the policy if a particular country adopts it, there is no guarantee that the IV estimator comes any closer to the desired target than the OLS estimator and indeed it may be more biased than OLS. Because different instruments define different parameters, having a wealth of different strong instruments does not improve the precision of the estimate of any particular parameter. This is in stark contrast with the traditional model with β ⊥ D. In that case, all valid instruments identify β̄. The Durbin (1954) – Wu (1973) – Hausman (1978) test for the validity of extra instruments applies to the traditional model....
[...]