Absence, Substitutability and Productivity: Evidence from Teachers Asma Benhenda Paris School of Economics December 2016 Still Very Preliminary 1 / 45
Motivation (1/2) • Worker absence : frequent in many countries ⇒ 2 to 3 % of annual work time (US, UK & France) • Empirical evidence on the causal effect of worker absence on productivity is scarce (Clotfelter et al., 2009 ; Duflo et al., 2012 ; Herrmann and Rockoff, 2012) • Even much less is known on organizations’ strategies to cope with this disruptive event • Research Question : • When a worker is absent, how does it hurt her productivity ? • How easily do organizations manage to mitigate this effect with substitute workers ? 2 / 45
Motivation (2/2) • The aim of this paper is : • to estimate the effect of teacher absence on student outcomes • to analyze substitute teachers’ assignment policies both across schools and within school, across classrooms • to study how the effect of teacher absences can be mitigated by the assignment and quality of substitute teachers • Important questions because : • Impact of worker health and effort on productivity • Specific human capital and its relationship with worker substituability • Teachers : consequences on educational inequality 3 / 45
Setting • Two types of absences : • One year or more : systematically replaced • Shorter term absences : not systematically replaced Two main types of substitute teachers : • • Tenured teachers assigned to a ZIP code area ( Titulaires sur Zone de Remplacement ) • Contract teachers hired on the spot, not trained nor certified 4 / 45
Preliminary Results (1/2) • One year substitute teachers’ assignment (9th grade) : • Across schools : Contract teachers are more likely to be assigned in disadvantaged regions and in disadvantaged schools • Within schools, across classrooms : around 85 % of schools match their substitute teachers to students depending on students’ socioeconomic status • One year absences (9th grade) : Contract teacher rather than a regular teacher • ⇒ decreases student test scores by 9 % of a standard deviation • Tenured sub. teacher rather than a regular teacher ⇒ decreases student test scores by 1 % of a standard deviation 5 / 45
Preliminary Results (2/2) Shorter term absences (9th grade) : • 10 additional days of absence ⇒ decreases student test scores by around 0.5 % of a standard deviation • Large heterogeneity by teaching subject : Math teachers’ absences have the largest effect 6 / 45
Data and Empirical Strategy : • Data : Administrative data matching each teacher to her student (2004-2015) : Focus on 9th grade Math, French and History teachers and their students Empirical Approaches : • • Substitute teachers assignment within school : test, for each school, whether the assignment of a substitute teachers predicts classroom characteristics (Horv´ ath, 2015) • One year absences : Within school, across cohort comparison = event study (Chetty et al., 2014) • Shorter term absences : Within teacher, across time comparison = teacher fixed-effect (Herrmann and Rockoff, 2012) • All absences : Within student, across topics comparison = student fixed-effect (Clotfelter et al., 2009) 7 / 45
Related Literature • Effect of worker absence on productivity (Miller et al., 2008 ; Clotfelter et al., 2009 ; Duflo et al., 2012 ; Herrmann and Rockoff, 2012) ⇒ substitute teachers • Classroom assignment policies (Horv´ ath, 2015) ⇒ substitute teachers’ assignment • Worker substituability (Hensvik and Rosenqvist, 2016 ; J¨ ager, 2016) ⇒ actual output , student outcomes ⇒ teaching = based on personal interactions = specific human capital (Ost, 2014) • Instruction time (Pischke, 2007 ; Lavy, 2015) ⇒ variations in the actual , rather than theoretical, amount of instruction hours 8 / 45
Conceptual Framework Data and Descriptive Statistics One Year Substitute Teachers Assignment Across Schools One Year Substitute Teachers Assignment Within School, Across Classrooms One Year Absences Short Term Absences Short Term Absences : Heterogeneity by Topic Conclusion 9 / 45
Average Hourly Productivity • Consider the average hourly productivity q j , i of teacher j over her hours of teaching with student i , indexed from 1 to T j , i : T j , i 1 � q j , i = q j , i , t (1) T j , i t =1 Crucial assumption : • q j , i = q j ( T j , i ) , with ∂ q j , i ( T j , i ) > 0 (2) ∂ T j , i ⇒ Intuition = teachers acquire, over their hours of teaching, student-specific human capital which contributes positively to their average productivity 10 / 45
Total Productivity • Total productivity Q T j , i over hours of teaching indexed from 1 to T j , i writes : Q T j , i = f T j , i ( q j , i , 1 , q j , i , 2 , ..., q j , i , T j , i ) , (3) � r if the regular teacher is teaching where j = s if the substitute teacher s is teaching 11 / 45
Student Output • From student i perspective, the total number of planned hours of instruction T i writes : T i = T i , r + T i , s + T i , a (4) where T i , a is the number of instruction hours lost by student i when her regular teacher is absent and no substitute teacher is assigned. • Y i , T i , student i output over T i writes : Y i , T i = g T ( f T i , r + f T i , s , T i , a , ǫ i , T i ) (5) • Assuming f T i , j and g T to be additive and separable : Y i , T i = T i , r q r ( T i , r ) .α + T i , s q s ( T i , s ) .β + T i , a .γ + ǫ i , T i (6) 12 / 45
Case 1 • Case 1 : Regular teacher is absent and a substitute teacher is assigned : ∂ Y i , T i = − α [ q r ( T i − T i , s ) + ∂ q r ( T i − T i , s ) ( T i − T i , s )] ∂ T i , s ∂ T i , s + β [ q s ( T i , s ) + ∂ q s ( T i , s ) T i , s ] ∂ T i , s Predictions : • 1 The more productive the regular teacher, the greater the output loss from her absence 2 The more productive the substitute teacher, the smaller the negative effect of absence 3 The effect of absence depends on its disruptive dimension and how fast teachers gain student-specific human capital 13 / 45
Case 2 • Case 2 : Regular teacher is absent and no substitute teacher is assigned : = − α [ q r ( T i − T i , a ) + ∂ q r ( T i − T i , a ) ∂ Y i , T i ( T i − T i , a )] + γ ∂ T i , a ∂ T i , a • Main prediction : ∂ Y i , T i > 0 ⇔ γ > α [ q r ( T i − T i , a ) + δ q r ( T i − T i , a ) ( T i − T i , a )] ∂ T i , a δ T i , a ⇒ students use their lost instruction hours so efficiently that these hours are more productive than the instruction hours they would have had with their missing teacher 14 / 45
Case 3 • Case 3 : Regular teacher is absent for the whole year and a substitute teacher is assigned : • Marginal effect of absence : δ Y i , T i δ q s ( T i ) = β [ q s ( T i ) + T i ] δ T i δ T i • Counterfactual : α [ q r ( T i ) + δ q r ( T i ) ( T i )] δ T i • Predictions : 1 Effect depends on the difference in quality between the regular and the substitute teachers 2 Effect depends on the substitute teacher student-specific human capital 15 / 45
Conceptual Framework Data and Descriptive Statistics One Year Substitute Teachers Assignment Across Schools One Year Substitute Teachers Assignment Within School, Across Classrooms One Year Absences Short Term Absences Short Term Absences : Heterogeneity by Topic Conclusion 16 / 45
Substitute Teachers Characteristics (2004 -2015) Tenured Sub. Contract Teacher Non substitute A. Demographics Male 0.39 (0.49) 0.41 (0.49) 0.36 (0.48) Age 35.23 (8.83) 37.16 (8.93) 43.41 (10.45) B. Certification 0.05 (0.22) 0 0.05 (0.22) Agr´ egation CAPES 0.69 (0.46) 0 0.76 (0.42) Other 0.25 (0.43) 0 0.18 (0.38) C. Teaching Subject Math 0.08 (0.28) 0.14 (0.34) 0.13 (0.33) French 0.16 (0.37) 0.16 (0.36) 0.18 (0.38) Other 0.76 (0.42) 0.70 (0.45) 0.69 (0.46) N 174,477 87,493 1,849,937 17 / 45
Proportion of Substitute Teachers (2004 - 2015) 0.10 0.09 0.08 Proportion of teachers 0.07 0.06 0.05 0.04 0.03 Contract Teacher 0.02 Tenured Sub. Teacher 0.01 2004 2006 2008 2010 2012 2014 18 / 45
Conceptual Framework Data and Descriptive Statistics One Year Substitute Teachers Assignment Across Schools One Year Substitute Teachers Assignment Within School, Across Classrooms One Year Absences Short Term Absences Short Term Absences : Heterogeneity by Topic Conclusion 19 / 45
Proportion of Substitute Teachers by Region in 2015 CRETEIL PARIS NANTES BORDEAUX NANCY METZ VERSAILLES CLERMONT … AIX MARSEILLE ORLEANS TOURS RENNES MONTPELLIER DIJON POITIERS NICE LIMOGES GRENOBLE LILLE LYON STRASBOURG ROUEN REIMS TOULOUSE AMIENS Tenured Sub. Teacher CAEN Contract Teacher BESANCON 0.00 0.02 0.04 0.06 0.08 0.10 0.12 20 / 45
Proportion of Substitute Teachers by School Percentile Rank at the 9th grade exam 0.12 Tenured Substitute Teacher 0.11 Contract Teacher Proportion of substitute teachers 0.1 0.09 0.08 0.07 0.06 0.05 0.04 0.03 0.02 0 20 40 60 80 100 School percentile rank at the 9th grade exam 21 / 45
Conceptual Framework Data and Descriptive Statistics One Year Substitute Teachers Assignment Across Schools One Year Substitute Teachers Assignment Within School, Across Classrooms One Year Absences Short Term Absences Short Term Absences : Heterogeneity by Topic Conclusion 22 / 45
Recommend
More recommend
