Lavy (Maks) (correct order) Flashcards
(120 cards)
1
Q
Title: Altonji, Bharadwaj & Lange (2012)
A
Changes in the Characteristics of American Youth: Implications for Adult Outcomes
2
Q
Motivation: Altonji, Bharadwaj & Lange (2012) - Changes in the Characteristics of American Youth: Implications for Adult Outcomes
A
Ask whether 1990s teens are really more “work-ready” than 1970s teens and how that could shape future wage gaps
3
Q
Setting: Altonji, Bharadwaj & Lange (2012) - Changes in the Characteristics of American Youth: Implications for Adult Outcomes
A
Two U.S. youth panels — one starting in the late-1970s, one in the late-1990s — tracking family background, AFQT scores, schooling, and actual or forecast adult wages
4
Q
Methodology: Altonji, Bharadwaj & Lange (2012) - Changes in the Characteristics of American Youth: Implications for Adult Outcomes
A
Re-weight 1970s cohort’s mix of race, parental education, AFQT, etc. to match the 1990s group, then feed into the 1970s wage equation. Then, add blocks of traits one at a time (parent ed. → test scores → own schooling → work timing) to see which shifts move predicted wages most.
5
Q
Robustness: Altonji, Bharadwaj & Lange (2012) - Changes in the Characteristics of American Youth: Implications for Adult Outcomes
A
Repeat with simple OLS/probit models; findings hardly move.
6
Q
Results: Altonji, Bharadwaj & Lange (2012) - Changes in the Characteristics of American Youth: Implications for Adult Outcomes
A
Skill-adjusted wages rise only a little; gains bigger at the top. Minorities and women narrow gaps (mainly via better-educated parents). Without faster skill growth, wage inequality is likely to keep widening.
7
Q
Title: Abramitzky & Lavy (2014)
A
How Responsive is Investment in Schooling to Changes in Returns? Evidence from an Unusual Pay Reform in Israel’s Kibbutzim
8
Q
Motivation: Abramitzky and Lavy (2014) - How Responsive is Investment in Schooling to Changes in Returns? Evidence from an Unusual Pay Reform in Israel’s Kibbutzim
A
Theory says students work harder in school when the payoff to education rises, yet clear real-world tests are rare
9
Q
Setting: Abramitzky and Lavy (2014) - How Responsive is Investment in Schooling to Changes in Returns? Evidence from an Unusual Pay Reform in Israel’s Kibbutzim
A
Israeli kibbutzim before vs. after a switch from equal pay to individual, performance-based pay. Two waves of 10th-graders tracked in admin records for school completion and exam results
10
Q
Methodology: Abramitzky and Lavy (2014) - How Responsive is Investment in Schooling to Changes in Returns? Evidence from an Unusual Pay Reform in Israel’s Kibbutzim
A
Difference-in-Differences: Early-reform kibbutzim are “treated,” late-reform ones are “controls.” Compare how student outcomes change over time across the two groups, checking that pre-trends match and that effects grow with longer exposure
11
Q
Robustness: Abramitzky and Lavy (2014) - How Responsive is Investment in Schooling to Changes in Returns? Evidence from an Unusual Pay Reform in Israel’s Kibbutzim
A
Parallel-trend and balance tests, very little in- or out-migration
12
Q
Results: Abramitzky and Lavy (2014) - How Responsive is Investment in Schooling to Changes in Returns? Evidence from an Unusual Pay Reform in Israel’s Kibbutzim
A
After pay reform, high-school completion, test scores, and both basic and college-track diplomas all move up. Gains are bigger for students from less-educated parent backgrounds, stronger for boys, effect rises with more years under the performance based system.
13
Q
Title: Abramitzky, Lavy and Segev (2024)
A
The Effect of Changes in the Skill Premium on College Degree Attainment and the Choice of Major
14
Q
Motivation: Abramitzky, Lavy and Segev (2024) - The Effect of Changes in the Skill Premium on College Degree Attainment and the Choice of Major
A
When the payoff to a degree suddenly jumps, do young adults finish college more often and steer toward better-paid majors?
15
Q
Setting: Abramitzky, Lavy and Segev (2024) - The Effect of Changes in the Skill Premium on College Degree Attainment and the Choice of Major
A
Israeli kibbutzim before vs. after a switch from equal pay to individual, performance-based pay. Records track high-school ability, BA completion, chosen major (STEM vs Humanities, etc.) and later wages.
16
Q
Methodology: Abramitzky, Lavy and Segev (2024) - The Effect of Changes in the Skill Premium on College Degree Attainment and the Choice of Major
A
Difference-in-Differences: compare early-reform youths with late-reform youths before vs after the switch, using kibbutz and cohort fixed effects. Split outcomes by major type to see if students re-sort toward high-return fields.
17
Q
Robustness: Abramitzky, Lavy and Segev (2024) - The Effect of Changes in the Skill Premium on College Degree Attainment and the Choice of Major
A
Parallel pre-trends, balanced backgrounds, placebo tests on non-treated outcomes, external comparison with Tel-Aviv youth.
18
Q
Results: Abramitzky, Lavy and Segev (2024) - The Effect of Changes in the Skill Premium on College Degree Attainment and the Choice of Major
A
After pay reform, BA completion rises sharply, with effect driven almost entirely by STEM fields; low-return majors stay flat. Men flock to computing/engineering, women to bio/chem and computing. Gains are biggest for students who already met advanced-math prerequisites. Expected and realised wages move up.
(Peopple flock to high return degrees)
19
Q
Title: Jensen (2010)
A
The Perceived Returns to Education and the Demand for Schooling
20
Q
Motivation: Jensen (2010) - The (Perceived) Returns to Education and the Demand for Schooling
A
Students often drop out after primary even though finishing secondary pays a lot more. The study asks: (i) How wrong are students’ wage beliefs? (ii) If you simply tell them accurate returns, do they stay in school?
21
Q
Setting: Jensen (2010) - The (Perceived) Returns to Education and the Demand for Schooling
A
8th-grade students in public schools spread across Dominican Republic — capturing pupils right before the stay/leave decision for Grade 9
22
Q
Methodology: Jensen (2010) - The (Perceived) Returns to Education and the Demand for Schooling
A
RCT: half the schools hear a speaker read official stats showing secondary school grads earn much more; the other half hear nothing. Difference-in-difference within each student a few months later for belief shifts, then track schooling for four years.
23
Q
Robustness: Jensen (2010) - The (Perceived) Returns to Education and the Demand for Schooling
A
Results survive alternative controls and comparison with non-surveyed classmates. Randomization balances backgrounds.
24
Q
Results: Jensen (2010) - The (Perceived) Returns to Education and the Demand for Schooling
A
Beliefs jump sharply — students now expect far bigger wage gains. Enrollment in Grade 9 increases, and total schooling inches higher, but mainly among boys from better-off families; poorer students revise beliefs yet do not stay longer, pointing to credit constraints
25
Title: Bertrand & Mullainathan (2004)
Are Emily and Greg More Employable than Lakisha and Jamal? — A Field Experiment on Labor-Market Discrimination
26
Motivation: Bertrand and Mullainathan (2004) - Are Emily and Greg More Employable than Lakisha and Jamal? A Field Experiment on Labour Market Discrimination
Despite equal paper credentials, Black job-seekers face lower pay and higher unemployment. The authors test whether race cues alone sways employers at the very first screen
27
Setting: Bertrand and Mullainathan (2004) - Are Emily and Greg More Employable than Lakisha and Jamal? A Field Experiment on Labour Market Discrimination
Resumes were randomly given either a White-sounding or a Black-sounding first name and mailed to job ads in Boston and Chicago. Researchers tracked whether firms responded back.
28
Methodology: Bertrand and Mullainathan (2004) - Are Emily and Greg More Employable than Lakisha and Jamal? A Field Experiment on Labour Market Discrimination
RCT: Within each ad, the only thing that changes across otherwise identical CVs is the first name’s race signal (plus, independently, résumé quality). Regressing Callback likelihood on Black, credentials, Black x Credentials variables, adding ad, city and occupation controls.
29
Robustness: Bertrand and Mullainathan (2004) - Are Emily and Greg More Employable than Lakisha and Jamal? A Field Experiment on Labour Market Discrimination
Gaps look the same across both cities, job types, good and bad neighbourhoods, and among firms that claim to be equal-opportunity employers
30
Results: Bertrand and Mullainathan (2004) - Are Emily and Greg More Employable than Lakisha and Jamal? A Field Experiment on Labour Market Discrimination
Resumes with White-sounding names get substantially more interview calls than identical resumes with Black-sounding names. Improving credentials boosts callbacks mainly for White names.
31
Title: Lavy (2008)
Do Gender Stereotypes Reduce Girls’ or Boys’ Human-Capital Outcomes? Evidence from Teacher Grading Bias
32
Motivation: Lavy (2008) - Do Gender Stereotypes Reduce Girls’ or Boys’ Human Capital Outcomes? Evidence from a Natural Experiment
Teachers are thought to grade girls and boys differently, but hard proof is scarce. Do marks shift once the grader can infer gender?
33
Setting: Lavy (2008) - Do Gender Stereotypes Reduce Girls’ or Boys’ Human Capital Outcomes? Evidence from a Natural Experiment
Israel’s matriculation system allows a clean test: every student gets one blind state exam mark and one non-blind school mark. Hundreds of thousands of paired blind vs. non-blind scores across 9 subjects.
34
Methodology: Lavy (2008) - Do Gender Stereotypes Reduce Girls’ or Boys’ Human Capital Outcomes? Evidence from a Natural Experiment
Within-pupil DiD: compare each child’s non-blind score with their own blind score, then see whether that gap is different for boys and girls across subjects.
35
Robustness: Lavy (2008) - Do Gender Stereotypes Reduce Girls’ or Boys’ Human Capital Outcomes? Evidence from a Natural Experiment
Add past scores and parent education, split by ability quartile, school type is not important and timing is checked, using later tests. DiD removes teacher inflation, SEs clustered at class level.
36
Results: Lavy (2008) - Do Gender Stereotypes Reduce Girls’ or Boys’ Human Capital Outcomes? Evidence from a Natural Experiment
When teachers know the name, boys’ marks fall relative to blind grading in almost every subject (especially languages and science labs). The penalty shows up for high and low achievers alike, is stronger under male or younger graders.
37
Title: Alesina et al. (2024)
Revealing Stereotypes: Evidence from Immigrants in Schools
38
Motivation: Alesina et al. (2024) - Revealing Stereotypes: Evidence from Immigrants in Schools
Immigrant pupils often get lower report-card marks than natives who score the same on blind tests. Is that gap linked to teachers’ hidden anti-immigrant bias? And can simple feedback make grading fairer?
39
Setting: Alesina et al. (2024) - Revealing Stereotypes: Evidence from Immigrants in Schools
Students' marks from Italian middle schools, linked to each pupil’s blind INVALSI test score. Teachers also took a 5-minute Implicit-Association Test (IAT).
40
Methodology: Alesina et al. (2024) - Revealing Stereotypes: Evidence from Immigrants in Schools
(i) Compare student’s teacher mark minus blind score and see if the gap is larger for immigrants. (ii) Treatment teachers saw their own IAT score 1 week before grading; control teachers saw it 1 week after. Measure how the immigrant-native grade gap changes.
41
Robustness: Alesina et al. (2024) - Revealing Stereotypes: Evidence from Immigrants in Schools
Pupil & class controls, placebo re-labels, parallel-trend tests; online grading exercise replicates patterns.
42
Results: Alesina et al. (2024) - Revealing Stereotypes: Evidence from Immigrants in Schools
Immigrant pupils get lower marks, and the penalty grows with the teacher’s bias score. Seeing one’s own IAT number roughly halves the gap — mostly among the most biased graders.
43
Title: Hanna & Linden (2012)
Discrimination in Grading
44
Motivation: Hanna and Linden (2012) - Discrimination in Grading
Check if teachers unconsciously mark identical work lower when they think it comes from a low-caste child, reinforcing inequality.
45
Setting: Hanna and Linden (2012) - Discrimination in Grading
An exam contest in an Indian city, where children sat maths, language and art papers.
46
Methodology: Hanna and Linden (2012) - Discrimination in Grading
RCT: treatment scripts carry a low-caste name. Identical control scripts carry a high-caste name. Random assignment inside each teacher’s pile means any score gap is pure grading bias. Teacher fixed-effects complement regressions.
47
Robustness: Hanna and Linden (2012) - Discrimination in Grading
Balance checks show labels unrelated to blind marks; adding pupil covariates or blind scores leaves the caste penalty unchanged; placebo relabellings and alternative clustering give the same result. Grader fixed-effects.
48
Results: Hanna and Linden (2012) - Discrimination in Grading
Scripts tagged low-caste receive slightly lower marks on average. The difference is largest in the first few papers graded and disappears as they see more scripts, suggesting statistical discrimination under greater uncertainty. With the penalty driven by low-caste graders to low-caste students.
49
Title: Lavy & Sand (2018)
On the Origins of Gender Gaps in Human Capital: Short- and Long-Term Consequences of Teachers’ Biases
50
Motivation: Lavy and Sand (2018) - On The Origins of the Gender Human Capital Gap: Short and Long Term Effect of Teachers’ Stereotypes
Ask whether tiny, often hidden, pro-boy marks that primary-school teachers give can snowball into the big gender gaps in human capital (later gades & advanced course choice)
51
Setting: Lavy and Sand (2018) - On The Origins of the Gender Human Capital Gap: Short and Long Term Effect of Teachers’ Stereotypes
Tel-Aviv pupils first sit a blind national maths test in Grade 5, then an almost identical non-blind school test in Grade 6 that their own teacher marks.
52
Methodology: Lavy and Sand (2018) - On The Origins of the Gender Human Capital Gap: Short and Long Term Effect of Teachers’ Stereotypes
1) Build a teacher-bias index by comparing, within each class, the blind & teacher marked test for boys and then for girls (see direction of bias).
2) Use within-school regressions to regress future outcomes (future scores/choice of stem couces) on the corresponding teacher's bias, while controlling for prior achievement and fixed school factors.
53
Robustness: Lavy and Sand (2018) - On The Origins of the Gender Human Capital Gap: Short and Long Term Effect of Teachers’ Stereotypes
Jack-knife versions of the bias index, pupil covariates, placebo tests, class and subject fixed effects, and balance checks all leave the pattern intact
54
Results: Lavy and Sand (2018) - On The Origins of the Gender Human Capital Gap: Short and Long Term Effect of Teachers’ Stereotypes
A small tilt that inflates boys’ maths marks in Grade 6 raises their later test scores and steers them into STEM tracks, while the same tilt holds girls back. The advantage persists for years. Removing the average pro-boy bias would close a noticeable share of the gender gap.
55
Title: Nunn & Qian (2011)
The Potato’s Contribution to World Population and Urbanization
56
Motivation: Nunn and Qian (2011) - The Potato's Contribution to Population and Urbanization: Evidence from a Historical Experiment
Global population and city growth surged after the 1700s. The authors ask whether the arrival of the potato—an energy-dense New-World crop—was a key causal driver.
57
Setting: Nunn and Qian (2011) - The Potato's Contribution to Population and Urbanization: Evidence from a Historical Experiment
Country-level panel from the Old World, 1000-1900: maps showing how much of each country could grow potatoes, matched to long-run population counts
58
Methodology: Nunn and Qian (2011) - The Potato's Contribution to Population and Urbanization: Evidence from a Historical Experiment
Continuous “dose” DiD: regress urbanization/population outcome on interaction term potato-suitable-land (a climate-and-soil trait that never changes) * post 1700 (time after potato adption), alongside country and time fixed effects and controls.
The coefficient asks: after potatoes became available, did countries that happened to have more potato-friendly land grow faster than those that did not?
59
Robustness: Nunn and Qian (2011) - The Potato's Contribution to Population and Urbanization: Evidence from a Historical Experiment
1) placebo with potato arrival: rerun the same difference-in-differences but pretend the potato arrived early, labelling pre-1700 windows as the “post” period; the interaction between potato-suitable land and this fake post dummy is always near zero and insignificant, showing that potato-friendly geography only boosts growth once the crop is actually available.
2) controls for countries andtime (geography, disease shocks, institutions ect)
60
Results: Nunn and Qian (2011) - The Potato's Contribution to Population and Urbanization: Evidence from a Historical Experiment
Countries that could grow potatoes saw faster population increases and urban expansion after 1700. Timing matches the crop’s diffusion, and supporting evidence on height and city growth points to better nutrition and food surplus as the mechanism.
61
Title: Jayachandran & Pande (2017)
Why Are Indian Children So Short? The Role of Birth Order and Son Preference
62
Motivation: Jayachandran and Pande (2017) - Why Are Indian Children So Short? The Role of Birth Order and Son Preference
Indian children are shorter than peers in poorer countries. The study asks whether unequal treatment of later-borns, fueled by a cultural bias toward eldest sons, can explain this puzzle.
63
Setting: Jayachandran and Pande (2017) - Why Are Indian Children So Short? The Role of Birth Order and Son Preference
National health surveys covering hundreds of thousands of children across India and many African countries provide height-for-age scores
64
Methodology: Jayachandran and Pande (2017) - Why Are Indian Children So Short? The Role of Birth Order and Son Preference
1) Regress height-for-age (z-scores) on birth-order dummies, an India dummy, and the interactions India × 2nd-child and India × 3rd-ect. negative interaction coefficients show later-born Indian kids lose extra height relative to later-born African kids
2) Adding neighbourhood, mother, and child fixed effects, limiting the sample to mothers past child-bearing, and replacing Africa with other country groups—confirming the extra penalty survives every “apples-to-apples” check.
3) To trace the mechanism, they looked at
i) Indian groups with weaker son preference (muslim,kerala),
ii) whether the penalty hinges on the gender mix of older siblings,
iii) compare prenatal and post-natal inputs that taper off for later-borns;
the consistent patterns point to parental eldest-son bias, not biology or poverty, as the source of the gap.
65
Robustness: Jayachandran and Pande (2017) - Why Are Indian Children So Short? The Role of Birth Order and Son Preference
Alternative country benchmarks keep the steep Indian gradient intact. Add mother fixed effects to contrast siblings within the same family, so any extra Indian penalty reflects parental behaviour, not poverty or genetics.
66
Results: Jayachandran and Pande (2017) - Why Are Indian Children So Short? The Role of Birth Order and Son Preference
1) Indian first-borns are just as tall as their African peers: short height only kicks in for later-born kids and deepens with each extra birth order.
2) post/pre natal care: Drop in check-ups, supplements, safe deliveries and vaccinations are much steeper across birth orders in India than in Africa, mirroring the height pattern.
3) Culture is the mechanism: groups with weaker eldest-son preference: Muslims, matrilineal Kerala show a flatter birth-order penalty.
4) Sibling gender: A second-born boy thrives only when his older sibling is a girl (making him the first son), while late sibling girls fare worst, confirming an eldest-son bias in parental investment (dont invest until there is a son). Shorter than african girls
5) Girls lose overall: Indian girls are disadvantaged from the very first birth and fall further behind with each subsequent birth: "keep-trying-for-a-son"
67
Title: Acemoglu, Johnson & Robinson (2001)
The Colonial Origins of Comparative Development: An Empirical Investigation
68
Motivation: Acemoglu, Johnson & Robinson (2001) – The Colonial Origins of Comparative Development: An Empirical Investigation
Why are some countries much richer than others? Georgraphy, culture, colonial identity leave much unexplained.
Institutions are a good explanations, but they are endogenous - settler mortality (showing where insitutions began & property rights established) is a good proxy.
(FYI property rights = institutions = institutins index)
69
Data/Setting: Acemoglu, Johnson & Robinson (2001) – The Colonial Origins of Comparative Development: An Empirical Investigation
Cross-country dataset links eighteenth-century European settler mortality figures to nineteenth-century measures of institutional checks on elites and to late-twentieth-century income per person for a broad sample of former colonies spread across all continents.
70
Methodology: Acemoglu, Johnson & Robinson (2001) – The Colonial Origins of Comparative Development: An Empirical Investigation
Institutions (property rights is the institution score) are endogenous, so settler mortality is used as an IV instrument.
1) First stage: regress institutional quality on settler mortality (relevance check) -> compute predicted institutional score (instituinal score stripped of endogeneity and is predicted only by settler mortality +contols)
2) second stage: regress modern GDP per capita on predicted institutions, adding continent and geography controls in a two-stage least-squares framework.
assumption: (rellevance) settler mortality has effect on type of institution (settler/vs extractive), but has no direct effect on 1995 income (deaths long ago dont effect today's gdp)
settler mortality -> where insitutions are estabished (extraction or settlment) -> better long term growth.
71
Robustness Checks: Acemoglu, Johnson & Robinson (2001) – The Colonial Origins of Comparative Development: An Empirical Investigation
Estimates hold when excluding specific regions, adding 13 controls (climate ect), using alternative institutional indices, or clustering by coloniser. Pre-colonial prosperity measures and religion dummies do not overturn the instrumented institution–income link.
72
Results: Acemoglu, Johnson & Robinson (2001) – The Colonial Origins of Comparative Development: An Empirical Investigation
Settler-era mortality predicts today’s institutions (property ights): safer colonies kept inclusive rules; deadly ones kept extractive rules. Geography has limited predictive power
Mortality itself has no direct effect on income—its impact flows entirely through the institutions it shaped.
The evidence supports institutions as the deep driver of comparative development - stronger property rights could more than double GDP in today’s weakest-institution countries.
73
Title: Acemoglu, Johnson & Robinson (2002)
Reversal of Fortune: Geography and Institutions in the Making of the Modern World Income Distribution
74
Motivation: Acemoglu, Johnson & Robinson (2002) – Reversal of Fortune: Geography and Institutions in the Making of the Modern World Income Distribution
The authors notice that the places that were thriving when Europeans first arrived (dense cities, big empires e.g. inca) are today the poorest, and they want to know why.
The puzzle challenges “geography explains prosperity” stories and pushes the authors to ask whether the institutions Europeans implanted can account for this dramatic flip.
75
Data/Setting: Acemoglu, Johnson & Robinson (2002) – Reversal of Fortune: Geography and Institutions in the Making of the Modern World Income Distribution
Dataset pairs proxies for pre-colonial prosperity—urbanisation rates and population density around 1500—with modern GDP per person across territories colonised by European powers, complemented by the earlier settler-mortality and institution measures.
76
Methodology: Acemoglu, Johnson & Robinson (2002) – Reversal of Fortune: Geography and Institutions in the Making of the Modern World Income Distribution
1) simple regression shows comparison of modern gdp per capita with proxies for prosperity in 1500 (urbanisation/popn density). Negative relationship: the richer it was in 1500, the poorer it is today - Fortune reversal. Controlling for geography and coloniser effects.
2) observe that richer-then are poorer-now only where Europeans has colonial rule. show those same places inherited weaker property-rights (insitutions)
The negative slope disappears (turns positive) in places never colonised
3) Same IV method as 2001 - settler mortality -> colonial extraction (rich then) -> weak insitutions & poor today
77
Robustness Checks: Acemoglu, Johnson & Robinson (2002) – Reversal of Fortune: Geography and Institutions in the Making of the Modern World Income Distribution
1) Alternative data: swap in four other urbanisation series and a land-adjusted density measure—negative slope holds each time.
2) Instrument noise out: 2SLS using 1000 A.D. density → 1500 density (proxy for "richness") makes the negative slope larger, confirming it isn’t measurement error.
3)Colonial-only phenomenon: run the same regressions on regions never colonised—slope flips positive, proving geography alone can’t explain it.
78
Results: Acemoglu, Johnson & Robinson (2002) – Reversal of Fortune: Geography and Institutions in the Making of the Modern World Income Distribution
Regions that were densely populated and wealthy before colonisation now lag behind previously poor settler areas because colonisers installed extractive institutions in the former and settler ones in the latter, confirming that institutional paths, not inherent geography, drove the dramatic income flip.
79
Title: Imberman,et al (2012)
Katrina’s Children: A Natural Experiment in Peer Effects from Hurricane Evacuees
80
Motivation: Imberman, Kugler & Sacerdote (2012) – Katrina’s Children: A Natural Experiment in Peer Effects from Hurricane Evacuees
The hurricanes abruptly displaced ≈400,000 children, scattering them across host schools that had little say over how many or which evacuees they received (quasi-random allocation)
This random shock created a rare natural experiment: schools’ peer mix is quasi-random exogenous (no family influence/academic tackingg)
Authors measure how peers affect native students’ achievement and behaviour
81
Data/Setting: Imberman, Kugler & Sacerdote (2012) – Katrina’s Children: A Natural Experiment in Peer Effects from Hurricane Evacuees
. Louisianna houston panel, student records before/after hurricane, evacuee shares fo every school. Mainly black free llunch kids
82
Methodology: Imberman, Kugler & Sacerdote (2012) – Katrina’s Children: A Natural Experiment in Peer Effects from Hurricane Evacuees
Continious dose DID: regress school educational outcome (scoes,attendance,discipline) on fraction of school who are evacuees + school fixed effects & time fixed effects
Re-run the same model after slicing the Evacuee shares into quartiles/medians: evacuees’ own pre-Katrina achievement, revealing whether low-ability or high-ability newcomers drive the effects.
83
Robustness Checks: Imberman, Kugler & Sacerdote (2012) – Katrina’s Children: A Natural Experiment in Peer Effects from Hurricane Evacuees
1) Exogenous shock: Evacuee children were bused or stuck in shelters with almost no school choice, so the jump in each school’s evacuee share is quasi random once we hold school fixed effects constant (family couldn't chose/sort intoo schools). - Natives didn't leave either
2) Parallel-trend/Placebocheck: When the authors regress pre-Katrina outcomes on future evacuee shares, the coefficients are near zero—showing no hidden pre-trends. (future share couldn't have influenced prior scores )
3) IV (houston): Since some evacuees reshuffled after the first chaotic weeks in october, so they instrument the October evacuee share with the share before families could “shop” for schools. The first stage is strong, and 2SLS estimates are similar to OLS, supporting causal interpretation.
84
Results: Imberman, Kugler & Sacerdote (2012) – Katrina’s Children: A Natural Experiment in Peer Effects from Hurricane Evacuees
Average academic effects are small, but peer quality matters: high-achieving evacuees raise incumbents’ scores, low-achievers lower them.
Secondary schools receiving more evacuees experience higher absenteeism and discipline problems.
class size, spending and teacher quality stay steady, so the effects come from the kids, not crowded classrooms.
85
Title: Lavy & Sand (2019)
The Effect of Social Networks on Students’ Academic and Non-Cognitive Behavioral Outcomes: Evidence from Conditional Random Assignment of Friends in School
86
Motivation: Lavy & Sand (2019) – The Effect of Social Networks on Students’ Academic and Non-Cognitive Behavioral Outcomes: Evidence from Conditional Random Assignment of Friends in School
Tel Aviv’s middle-school choice process often splits sixth-grade friendship groups. The authors ask whether preserving or breaking specific friendship ties during this transition influences later test performance, happiness and behaviour.
87
Data/Setting: Lavy & Sand (2019) – The Effect of Social Networks on Students’ Academic and Non-Cognitive Behavioral Outcomes: Evidence from Conditional Random Assignment of Friends in School
Three cohorts of Tel Aviv pupils list up to eight desired friends when ranking middle-school options. These lists define “reciprocal,” “follower,” and “non-reciprocal” ties.
88
Methodology: Lavy & Sand (2019) – The Effect of Social Networks on Students’ Academic and Non-Cognitive Behavioral Outcomes: Evidence from Conditional Random Assignment of Friends in School
1) Children rank friends for schools, but the allocation within same class is random. Since allocation is random, the effect of keeping/breaking ties is causal.
2) OLS within school-cohort regresses eighth-grade test z-scores or behavioural indices on number of reciprocal, follower and non-reciprocal friends in class, controlling for those same friend counts in school plus class and year fixed effects.
89
Robustness Checks: Lavy & Sand (2019) – The Effect of Social Networks on Students’ Academic and Non-Cognitive Behavioral Outcomes: Evidence from Conditional Random Assignment of Friends in School
Chi-square tests confirm random class assignment: balancing shows friend counts uncorrelated with background traits.
Results persist with classroom fixed effects, added student covariates, alternative friend definitions and placebo checks using earlier cohorts.
90
Results: Lavy & Sand (2019) – The Effect of Social Networks on Students’ Academic and Non-Cognitive Behavioral Outcomes: Evidence from Conditional Random Assignment of Friends in School
Having more reciprocal friends boosts test scores (doubles for extra reciprocal). Followers also help exam scores but less. Non-reciprocal friends hurt performance a bit. Rejecters do nothing
More reciprocal friends mean less violence and greater social & school satisfaction
Academic payoff of "right group of friends" is bigger for boys
91
Title: Lavy & Schlosser (2011)
Mechanism and Impacts of Gender Peer Effects at School
92
Motivation: Lavy & Schlosser (2011) – Mechanism and Impacts of Gender Peer Effects at School
The study tackles whether the share of girls in mixed Israeli classrooms shapes learning, behaviour and classroom climate
93
Data/Setting: Lavy & Schlosser (2011) – Mechanism and Impacts of Gender Peer Effects at School
National Ministry datasets link every pupil’s background, test scores and matriculation records to grade-level gender ratios for multiple adjacent cohorts within each public school that retains mixed classes. Complementary student and teacher surveys report classroom climate and staff burnout.
94
Methodology: Lavy & Schlosser (2011) – Mechanism and Impacts of Gender Peer Effects at School
OLS: Regress scores and climate on the grade-level share of girls, controlling for student traits, grade & year dummies, and school fixed effects.
the share of girls is random: Law bans ability streaming, school gender share reflects andom birth-cohort luck. Grade level shares are taken to avoid streaming of bys & girls
95
Robustness Checks: Lavy & Schlosser (2011) – Mechanism and Impacts of Gender Peer Effects at School
1) Monte-Carlo simulations showing cohort-to-cohort girl-share shifts mimic random draws
2) balance tests: shifts in girl share are uncorrelated with shifts in parental education, ethnicity, immigration, cohort size, etc.
(iii) placebo cohorts give null effects: girl-share of the previous or next cohort in the same school: a mix that has no contact with the tested students as a fake treatment. When this substitute is plugged into the regression, the coefficients collapse to zero, showing that only the classmates sitting in your own grade drive the original effects, not school-wide shocks
96
Results: Lavy & Schlosser (2011) – Mechanism and Impacts of Gender Peer Effects at School
1) More girls in a grade lift test scores for boys and girls, with gains biggest when girls become the majority (grows non-linearly).
2) The classrooms calmer, less disruptions, friendlier and less-tired/burnt teachers. However own study effort doesn't change (thus class composition, not personal reform drives gains)
3) Effects are strongest for low-education families and recent immigrants.
97
Title: Gould, Lavy & Paserman (2004)
Immigrating to Opportunity: Estimating the Effect of School Quality Using a Natural Experiment on Ethiopians in Israel
98
Motivation: Gould, Lavy & Paserman (2004) – Immigrating to Opportunity: Estimating The Effect of School Quality Using a Natural Experiment on Ethiopians in Israel
Do Ethiopian children placed in better primary schools and richer community achieve better high school outcomes?
Operation Solomon airlift randomly scattered Ethiopian children across Israeli absorption centres (exogenous as bus routes and availability). Authors leverage this natural lottery to test whether the quality of the first elementary school influences later high-school persistence and matriculation success.
99
Data/Setting: Gould, Lavy & Paserman (2004) – Immigrating to Opportunity: Estimating The Effect of School Quality Using a Natural Experiment on Ethiopians in Israel
Administrative panels trace each Ethiopian pupil from initial elementary placement through high-school careers, linking them to pre-immigration school test averages and to community characteristics such as income and welfare rates.
school quality measured by average maths and hebrew scores (before arrival)
100
Methodology: Gould, Lavy & Paserman (2004) – Immigrating to Opportunity: Estimating The Effect of School Quality Using a Natural Experiment on Ethiopians in Israel
Probit and linear models: regress highschool outcomes on initial school math quality, controlling for age, sex, family size and parental education;
arrival-centre fixed effects absorb any locational sorting; sibling fixed-effect specifications purge family unobservables.
101
Robustness Checks: Gould, Lavy & Paserman (2004) – Immigrating to Opportunity: Estimating The Effect of School Quality Using a Natural Experiment on Ethiopians in Israel
1) Balance tests show that allocation was random (school quality uncorrelated with any pre-arrival family trait.)
2) Same street test: add locality fixed effects—effects unchanged.
3) control for fixed effects, different indeces, clustered SEs, results unchanged
102
Results: Gould, Lavy & Paserman (2004) – Immigrating to Opportunity: Estimating The Effect of School Quality Using a Natural Experiment on Ethiopians in Israel
1) Stay in school longer – Ethiopian children in bette schools far more likely to reach final years and to advance without repeating a year.
2) Exposure to a top-quartile school almost doubles the chance of sitting for – and passing – matriculation exam.
3) Better high-schools downstream – Those early placements also channel pupils into academically stronger secondary schools.
Academic, not neightbourhood is key: Only the school’s pre-1991 math score predicts these gains. verbal scores and community SES proxies do not, confirming the effect comes from instructional quality rather than richer surroundings or peer mix.
103
Title: Duflo & Hanna (2005)
Monitoring Works: Getting Teachers to Come to School
104
Motivation: Duflo & Hanna (2005) – Monitoring Works: Getting Teachers to Come to School
Teacher absence was nearly half of all school days in rural Indian “one-teacher” centres.
The authors test whether monitoring plus cash incetives can get teachers to show up and, if so, whether children actually learn more.
105
Data/Setting: Duflo & Hanna (2005) – Monitoring Works: Getting Teachers to Come to School
120 NGO schools in tribal Rajasthan; 60 randomly assigned to the camera contract, 60 kept fixed pay.
Monthly unannounced visits record attendance; pupils take independent Hindi-math tests before and after the school year.
106
Methodology: Duflo & Hanna (2005) – Monitoring Works: Getting Teachers to Come to School
Randomised Controlled Trial (RCT): treatment teachers’ pay rises or falls per day above/below 21 verified days; control teachers’ salary fixed.
Two eqs: teacher absence and child score regressed on treatment dummy + fixed effects
IV: treatment status used as IV for "observed presence" to quantify how each additional open day increases learning (to see if presence effects test scores)
107
Robustness Checks: Duflo & Hanna (2005) – Monitoring Works: Getting Teachers to Come to School
1) camera records match unannounced spot-checks, dismissing “photo-only” cheating;
2) treatment effect stable across different teacher types
108
Results: Duflo & Hanna (2005) – Monitoring Works: Getting Teachers to Come to School
Monitoring halves absence, eliminates extreme absence and delivers more teaching time.
After a year, pupil scores are noticeably higher and regular-school transition rises; no evidence teachers slackened in-class effort.
109
Title: Lavy (2009)
Performance Pay and Teachers’ Effort, Productivity and Grading Ethics
110
Motivation: Lavy (2009) – Performance Pay and Teachers’ Effort, Productivity and Grading Ethics
Israel piloted cash bonuses for high-school English and maths teachers ranked on exam pass rates and mean scores; study asks whether individual incentives lift real learning and alter teaching behaviour.
111
Data/Setting: Lavy (2009) – Performance Pay and Teachers’ Effort, Productivity and Grading Ethics
49 low performing schools, but 18 were treated by accident (slightly above 45cut-off). For DID take those 18 schools and find schools in 40-50% area that are similiar to the accidentally treated - those will be controls
112
Methodology: Lavy (2009) – Performance Pay and Teachers’ Effort, Productivity and Grading Ethics
Quasi-randomised treatment from measurement error: among schools near the 45 % pass-rate cutoff, some were mistakenly treated (not meant to have cut off, ministry made random error - satisfied as-good-as-random & uncorrelated with school stats)
DID with “randomised-treatment” subset before and after bonus: regress outcomes on treated, post bonus, and their product, alongside school fixed effects.
113
Robustness Checks: Lavy (2009) – Performance Pay and Teachers’ Effort, Productivity and Grading Ethics
1) Results replicated with regression-discontinuity around threshold (DID schools below and above 45)
2) full eligible-school DiD (all 48 programme schools, mistakes + real ones, and 50 eligible schools that wereuntreated dueto logistics)
alternative bandwidths and rich covariates; no rise in internal grading inflation or strategic pupil exclusion detected.
114
Results: Lavy (2009) – Performance Pay and Teachers’ Effort, Productivity and Grading Ethics
Incentives raise exam participation, conditional pass rates and mean scores; teacher diaries show more after-school tutoring, diversified pedagogy and sharper student feedback. External exam marking rules out score manipulation. Cheap to implement
115
Title: Duchini et al. (2023)
School Management Takeover, Leadership Change, and Personnel Policy
116
Motivation: Duchini et al. (2023) – School Management Takeover, Leadership Change, and Personnel Policy
England lets external “Sponsors” assume control of failing secondary schools. Paper probes whether such takeovers change leadership, teacher workforce and pay architecture in ways that could explain earlier pupil-achievement gains.
117
Data/Setting: Duchini et al. (2023) – School Management Takeover, Leadership Change, and Personnel Policy
A large panel links national payroll and census records for teachers to 600 Sponsor-led academy conversions (2005-2022) and to comparable schools that convert later, giving multi-year windows before and after takeover.
118
Methodology: Duchini et al. (2023) – School Management Takeover, Leadership Change, and Personnel Policy
Difference-in-Differences: treated schools observed from three years before to two years after conversion compared to "late converter" similiar schools (control). (both schools get converted, but 6 years apart - we never witness 2nd conversion)
Regressions of head-teacher change, exits, hires and pay dispersion on a takeover indicator including school and year fixed effects; robustness adds Local-Authority-year FE.
119
Robustness Checks: Duchini et al. (2023) – School Management Takeover, Leadership Change, and Personnel Policy
matched did cross-check: repeat did against never converted schools that matched characteristics.
Parallel-trend graphs, placebo leads, alternative control windows and re-weighting by pre-takeover traits yield similar coefficients; teacher-level models with individual fixed effects confirm pay-structure findings.
120
Results: Duchini et al. (2023) – School Management Takeover, Leadership Change, and Personnel Policy
Takeover doubles probability of a new, better-paid head drawn from outstanding schools; low-value teachers exit while high-quality recruits and Teach First entrants arrive; Sponsors scrap seniority scales, widening performance-related pay spread and coinciding with later GCSE gains.
