1 / 6

Gustavo Bobonis ECO4060

Comments on: “Network Effects and the Educational Attainment of Young Immigrants” by Florian Hoffmann. Gustavo Bobonis ECO4060. Summary. Question: Are there neighborhood-based peer effects for immigrants Interesting question: concern with rate of assimilation

judd
Download Presentation

Gustavo Bobonis ECO4060

An Image/Link below is provided (as is) to download presentation Download Policy: Content on the Website is provided to you AS IS for your information and personal use and may not be sold / licensed / shared on other websites without getting consent from its author. Content is provided to you AS IS for your information and personal use only. Download presentation by click this link. While downloading, if for some reason you are not able to download a presentation, the publisher may have deleted the file from their server. During download, if you can't get a presentation, the file might be deleted by the publisher.

E N D

Presentation Transcript


  1. Comments on:“Network Effects and the Educational Attainment of Young Immigrants”by Florian Hoffmann Gustavo Bobonis ECO4060

  2. Summary • Question: Are there neighborhood-based peer effects for immigrants • Interesting question: concern with rate of assimilation (Huntington, 2004; McKinnon and Parent, 2005) • Contribution: ID strategy that improves upon BLM to control for OVB • Group-region-cohort variation • Results: modest network effects • Moving from a location with no ‘peers’ to the average relative ‘share’ of peers increases individuals’ school attainment by 0.11-0.23 years (= α*3.06*12.4) • Distinction from previous literature: evidence of selection (movers vs. non-movers) • My comments: empirical framework, data & samples, potential threats to validity, discussion of results

  3. Empirical Framework and Lit Review • Manski (1993) – “The Reflection Problem” • Simultaneity problem: person A’s actions affect person B’s actions and vice versa. • Distinct from: • Correlated unobservables (e.g., common environmental shocks) • Endogenous group membership (e.g., selection into networks) • Endogenous and exogenous interactions cannot be identified, but can identify evidence of some type of peer or network effect • How to interpret BLM-model network effect α? • Reference: Moffitt (2001) – ID best-response function • Experimentally altering group membership • Partial population experiments – portion of individuals within a group are directly treated

  4. Data, Samples, and Reference Groups • Immigrants who speak their mother tongue at home • Mother tongue is important determinant of ethnic identity (Alba, 1990) • Endogenous? Sample selection based on an action affected by network effects (“pressures to conform or to distinguish oneself”)? • Recommendation: Use sample of all immigrants & use indicator for non-English language spoken at home as dep. variable. ‘Network’ effect on language use? • Immigrants who arrived when young or U.S. born (2nd/3rd generation?) • U.S. born (2nd or 3rd generation) might have different reference groups • Recommendation: Use only 1st generation immigrants? • Density and ‘quality’ of other groups do not affect schooling decision? • e.g., “conformity” or “need of differentiation” from natives/other migrants? • May bias estimates of α? Depends on: Cov(cajk, cal,k) <, >, or = 0, for each l ≠ j αl network effect of other groups * Group all Spanish-speakers and French-speakers together?

  5. Potential Threats to Validity • Focus on “group-region-cohort” variation ID strategy: • What are some of the main identifying assumptions? Examples: • μjt = 0 (no shocks to MSAs that affect own and peers’ school attainment or migration into or out of MSA; e.g., MSA ‘business or political cycles’) • μkt = 0 (no shocks to language groups that affect own and peers’ school attainment; e.g., changes in group-specific tastes) You can include both μjt and μkt fixed effects. • μjkt = 0 (no shocks to environment of language groups in particular MSAs that affect own and peers’ school attainment) Example: political economy-based hypothesis • Individuals in language groups with relatively high share of population and (high or low) education levels demand more access to or quality of public schools • Affects both own and peer school achievement? • Mechanism may be more important for high school • Interpretation of reduced-form coefficient, might include a school quality mechanism

  6. Discussion of Results • How do we know whether α is ‘economically’ (or sociologically?) significant? • One std. deviation increase in cajk for a given mean school attainment level? • Compare to previous estimates of network effects? • Main estimates of α from group-region-cohort variation (Table 7): • Are stable around: [0.002 (0.002) – 0.006 (0.003)] • Probably cannot reject that they’re significantly different • Effect for non-movers is 0.003 (0.002) (less selection for this group?) • Parental background variable subgroup: • Estimate is -0.004 (0.003) • But, parental background for sub-sample of individuals living with parents. Self-selected sample? • What is main network effect estimate for sub-sample when you exclude parental background control?

More Related