Course Hero Logo

Panels b to e alternative cohort ranges these panels

Course Hero uses AI to attempt to automatically extract content from documents to surface to you and others so you can study better, e.g., in search results, to enrich docs, and more. This preview shows page 75 - 77 out of 90 pages.

Panels B to E – Alternative Cohort RangesThese panels include two expanded samples, 1968 to 1992 (Panel B) and 1969 to 1989 (Panel C), and twomore restrictive ranges. In Panel D, one cohort from each end of the baseline sample is removed, yieldinga range from 1971 to 1987.This sample no longer includes any fully post-reform cohorts.The data arerestricted to 1972 in Panel E, the final fully pre-FPE cohort. Removing additional cohorts on the later endof the range would remove significant and necessary identifying variation; therefore, the 1987 cohort remainsthe cutoff on the upper end of the range in Panel E.Panel F – Matched 1984 Start AgesIntensity measures are constructed using starting age information from the 1984 census.While the pre-reform timing of these data are ideal, the administrative boundaries are not consistent between the 1984 andpost-1991 periods. Therefore, while there is starting age information contained in the 1984 census, the leveltwo administrative information does not match with the zones used in the study. To adjust the 1984 datato the 1994 geographical boundaries, shapefiles from each time period provided byMinnesota PopulationCenter and the Ethiopian Central Statistical Agency(2017) are overlaid, and new start values for the post-1991 boundaries are calculated as the weighted averages of the start age value from the 1984 area and theportion of the post-1991 zone that is made up of that 1984 area. Unfortunately, this requires an unrealisticassumption of a consistent distribution of population within geographic area, and introduces a significantamount of measurement error into the start age calculations.A.44
Panel G – 1994 Start AgesStart ages from the 1994 census are used to calculate all intensity measures. The timing of this survey isproblematic in the sense that the MTI implementation had already begun at this time. This along with anyanticipation of the forthcoming FPE program could alter the decision to enter school in 1994.Panel H – Three Part TrendThe district-specific linear trends are replaced with a set of district-specific trends that are allowed to changeslope at two points, in 1978 and in 1987. On time entrants are partially treated beginning with the 1978cohort, and fully treated beginning with the 1987 cohort.Panel I – Regional TrendsThe district-specific linear trend is replaced with a region-specific linear trend.Panel J – No TrendsAll trend variables are removed from the estimating equations.Panel K – Only Zones in All Rounds of the DHSData are restricted only to zones with observations in all three of the rounds of the DHS survey.Thisincludes 25 of 30 zones in the non-MTI regions, and 48 of the 60 zones throughout Ethiopia.Panel L – Zones with Fewer than 4,000 Organized Violence Deaths (1989 to 1991)Data from the four zones with the highest level of pre-independence violence, three of which are in thenon-MTI sample, are removed from the sample.These zones contain more than 75 percent of all deathsincluded in the data over this time period.

Upload your study docs or become a

Course Hero member to access this document

Upload your study docs or become a

Course Hero member to access this document

End of preview. Want to read all 90 pages?

Upload your study docs or become a

Course Hero member to access this document

Term
Summer
Professor
N/A
Tags

Newly uploaded documents

Show More

  • Left Quote Icon

    Student Picture

  • Left Quote Icon

    Student Picture

  • Left Quote Icon

    Student Picture