The Value of Smarter Teachers:
International Evidence on
Teacher Cognitive Skills and Student Performance
Eric A. Hanushek
Stanford University
Marc Piopiunik
Ifo Institute
Simon Wiederhold
KU Eichstätt-Ingolstadt & Ifo
Third International PIAAC Conference
Madrid (Spain), November 6–8, 2016
Huge Student Performance Gaps across Developed Countries
580
PISA 2012
SGP
JPN
500
NLD
ESTCAN
FIN
BEL POL
DEU
AUT
AUS
SVN CZE
NZL IRL
DNK
GBRFRA
NOR
ITA
ESP
SVK RUS
LTU
SWE USA
460
ISR
GRC
TUR
420
Math
540
KOR
CHL
420
460
500
Reading
Note: Learning progress during one school year about 40 points.
540
580
Motivation
• Potential reasons for international student performance gaps
• Parents’ education, learning culture, educational institutions...
• Until now, no internationally comparable measure on teacher quality
• Teacher salaries used as proxy for teacher quality
• In particular, no measure on teacher cognitive skills, one important
dimension of teacher quality
Overview
• Quantify for the first time internationally comparable teacher skills in
numeracy and literacy for 31 (mostly OECD) countries
• Combine these country-level measures of subject-specific teacher skills with
micro-level student performance in math and reading
• Use several econometric approaches to investigate the role of teacher
cognitive skills for international student performance gaps
• Investigate determinants of teacher cognitive skills
Teacher Cognitive Skills: Data
• Numeracy/literacy skills of teachers come from Programme for the
International Assessment of Adult Competencies (PIAAC)
• “PISA for adults”, initiated by OECD with 33 participating countries
• Round 1: 2011/2012 (24 countries); Round 2: 2014/2015 (9 countries)
• Tested individuals aged 16–65 years in numeracy and literacy
(nationally representative samples)
• At least 5,000 individuals per country
• Drop two countries (Cyprus did not participate in PISA; Indonesia is
Jakarta only)
Countries Participating in PIAAC (Round 1 and Round 2)
Numeracy - Sample Item (Level 5)
Teacher Sample
• Use 4-digit occupation code
• Teacher =
• Primary school teacher
• Secondary school teacher
• “Other teacher” (e.g. special ed teacher, language teacher)
• Exclude vocational school teachers since only very few PISA students
(15 year-olds) affected by those teachers
• Exclude pre-kindergarten teachers
Teacher Cognitive Skills
• Since we do not observe the subjects the teachers are teaching, we use both
numeracy and literacy skills of all PIAAC teachers
• Compute median numeracy/literacy skills for each country
• Standardize skills at country level to have mean 0 and std. dev. 1
• Sample sizes:
• From 106 (Chile) to 413 teachers (Denmark); 834 teachers (Canada)
• On average 207 teachers per country
• Validation check of new measures:
Position of teacher cognitive skills among all adults in national PIAAC
samples similar to that in other nationally representative datasets with much
larger samples (for USA and Germany) Details
310
FIN
300
JPN
AUT
DEU
BEL
SWE
CZE
SGP NORNLD
FRA
IRL
290
DNK
SVK
SVN
LTU
280
GRCESP
AUS
NZL
CAN
GBR
KOR
EST
USA
POL
270
ITA RUS
ISR
TUR
CHL
260
Numeracy skills teacher
320
330
Teacher Cognitive Skills
260
270
280
290
300
Literacy skills teacher
310
320
330
For comparison:
Skills of employed adults in Canada for three educational groups
310
Master
300
FIN
Bachelor
JPN
AUT
DEU
BEL
SWE
CZE
SGP NORNLD
FRA
290
LTU
280
GRCESP
AUS
NZL
IRL
DNK
SVK
SVN
CAN
GBR
KOR
EST
USA
270
POL
Post-sec.
TUR
CHL
260
Numeracy skills teacher
320
330
Teacher Cognitive Skills
260
ITA RUS
ISR
Post-sec.
270
280
Bachelor
290
300
Literacy skills teacher
Master
310
320
330
Position of Teacher Cognitive Skills in the Skill Distribution of College
Graduates
Numeracy
Finland
Japan
Germany
Belgium
Sweden
Czech Republic
Netherlands
Singapore
Norway
France
Austria
Australia
New Zealand
Ireland
Denmark
Slovak Republic
Slovenia
Canada
United Kingdom
Korea
Lithuania
Estonia
United States
Spain
Greece
Poland
Italy
Russian Federation
Israel
Turkey
Chile
200
250
300
350
Numeracy skills
Vertical bars indicate median cognitive skills of teachers in a country. Horizontal bars show the interval of cognitive skill levels of
all college graduates (including teachers) between the 25th and 75th percentile.
Position of Teacher Cognitive Skills in the Skill Distribution of College
Graduates: Selected Countries
Numeracy
Finland
Japan
Singapore
Austria
Denmark
Slovak Republic
Spain
Poland
260
280
300
Numeracy skills
320
340
Vertical bars indicate median cognitive skills of teachers in a country. Horizontal bars show the interval of cognitive skill levels of
all college graduates (including teachers) between the 25th and 75th percentile.
Position of Teacher Cognitive Skills in the Skill Distribution of College
Graduates
Literacy
Finland
Japan
Australia
New Zealand
Netherlands
Sweden
Canada
Norway
Belgium
Germany
United States
Ireland
Singapore
Czech Republic
United Kingdom
Korea
France
Estonia
Poland
Austria
Slovak Republic
Spain
Denmark
Slovenia
Greece
Russian Federation
Lithuania
Israel
Italy
Chile
Turkey
200
250
300
350
Literacy skills
Vertical bars indicate median cognitive skills of teachers in a country. Horizontal bars show the interval of cognitive skill levels of
all college graduates (including teachers) between the 25th and 75th percentile.
Position of Teacher Cognitive Skills in the Skill Distribution of College
Graduates: Selected Countries
Literacy
Finland
Japan
Singapore
Poland
Austria
Slovak Republic
Spain
Denmark
260
280
300
Literacy skills
320
340
Vertical bars indicate median cognitive skills of teachers in a country. Horizontal bars show the interval of cognitive skill levels of
all college graduates (including teachers) between the 25th and 75th percentile.
“The top-performing systems we studied recruit their teachers from the top
third of each cohort graduate from their school system” (McKinsey (2007))
Teacher numeracy skills
260
300
320
280
Numeracy
FIN
NLD
DNK
SVK
SGP
NOR
AUT
JPN
DEU
BEL
CZE SWE
FRA
AUS
NZL
SVN
LTU
USA
IRL
CAN
GBR
EST
KOR
GRC
ESP
POL
ITA
RUS
ISR
TUR
35
40
45
CHL
50
55
Teacher position collgrad distribution (w/o teachers)
Teacher literacy skills
320
300
280
Literacy
FIN
JPN
NZL
NLD
SWE
NOR
BEL
SVK
DNK
AUT
SVN
LTU
ISR
260
ITA
FRA
POL
DEU SGP
IRL
KOR
EST
USA
CZE
GBR
RUS
TUR
35
40
45
50
Teacher position collgrad distribution (w/o teachers)
ESP
AUS
CAN
GRC
CHL
55
60
Parent Cognitive Skills
• Compute proxies for numeracy/literacy skills of parents of PISA students to
account for intergenerational persistence in cognitive skills
→ separate impact of smart teachers from impact of smart parents
• Idea: Match numeracy/literacy skills of PIAAC adults to actual PISA parents
• Use PIAAC adults with children who are potential parents of PISA
students w.r.t. age (17–44 years old at birth of PISA students)
• Match cognitive skills based on three observable characteristics
• gender
• education (3 categories)
• number of books at home (6 categories)
• Use maximum skills of mother and father
Data on Students
• Student math and reading performance from Programme for International
Student Assessment (PISA)
• Tests nationally representative samples of 15-year-old students
• Use PISA waves 2009 and 2012
• Student cohorts in these two PISA waves largely taught by teacher
cohorts tested in PIAAC (2011–2015)
• Standardize math/reading test scores at the student level across all countries
(mean=0, SD=1)
.8
SGP
.4
JPN
CAN
EST
0
POL
ESP
USA
LTU
ITA
-.4
RUS
FIN
NLD
BEL
DEU
NZLAUS
SVNDNK AUT
FRA
IRL
NORCZE
GBR
SVK
SWE
GRC
ISR
TUR
-.8
Math performance student
KOR
CHL
-2
-1
0
Numeracy skills teacher
coef = .20897443, (robust) se = .03883733, t = 5.38
1
2
.8
.4
SGP
0
CAN
NZL
AUS
POL
IRL BEL NLD
EST
NOR
DEU
FRA USA
GBR
DNK
SWE
ITA
CZE
ESP
SVN
ISR GRC
AUT
LTU
SVK
RUS
-.4
TUR
JPN FIN
CHL
-.8
Reading performance student
KOR
-2
-1
0
Literacy skills teacher
coef = .17820177, (robust) se = .02062038, t = 8.64
1
2
Student Performance and Teacher Cognitive Skills (OLS)
Student Math Performance
Teacher cognitive skills
Student Reading Performance
(1)
(2)
(3)
(4)
(5)
(6)
0.209∗∗∗
0.173∗∗∗
0.145∗∗∗
0.178∗∗∗
0.102∗∗∗
0.092∗∗∗
(0.038)
(0.031)
(0.032)
(0.020)
(0.020)
(0.022)
Parent cognitive skills
0.044∗∗
0.015
(0.017)
(0.016)
Student characteristics
X
X
X
X
Parent characteristics
X
X
X
X
School characteristics
X
X
X
X
Country characteristics
X
X
X
X
490,818
490,818
490,818
490,818
Students
Countries
Adj. R2
490,818
31
31
31
31
31
31
0.04
0.29
0.29
0.03
0.30
0.30
∗
Control variables
490,818
p<0.10, ∗∗ p<0.05, ∗∗∗ p<0.01
Simulation Analysis: Raising Teacher Numeracy Skills to the Finnish Level
Chile
Turkey
Israel
Russian Federation
Italy
Poland
Greece
Spain
United States
Estonia
Lithuania
Korea
United Kingdom
Canada
Slovenia
Slovak Republic
Denmark
Ireland
New Zealand
Australia
Austria
France
Norway
Singapore
Netherlands
Czech Republic
Sweden
Belgium
Germany
Japan
0
.2
.4
.6
.8
1
1.2
Student performance increase (in school years)
1.4
Simulation Analysis: Raising Teacher Literacy Skills to the Finnish Level
Turkey
Chile
Italy
Israel
Lithuania
Russian Federation
Greece
Slovenia
Denmark
Spain
Slovak Republic
Austria
Poland
Estonia
France
Korea
United Kingdom
Czech Republic
Singapore
Ireland
United States
Germany
Belgium
Norway
Canada
Sweden
Netherlands
New Zealand
Australia
Japan
0
.2
.4
.6
.8
1
1.2
Student performance increase (in school years)
1.4
Robustness Checks
• Replace individual-level parent skills with country-level parent/adult skills
(same level as teacher skills) Details
• Add further country-level controls (e.g., teacher pay measures, institutional
measures) Details
• Add continental fixed effects or restrict the analysis to just Europe
Details
• Exclude each country individually from sample
• Teacher cognitive skills do not merely reflect pedagogical skills
• Add country-level instruction-practice indicators as proxy for teacher
pedagogical skills
• PISA asks students about the instructional practices of their teachers in
their reading (PISA 2009) and math (PISA 2012) classes
[4-point scale ranging from “never or rarely” to “almost or almost always”]
• Example: “Asking questions that make students reflect on the problem”
Items
Student Fixed-Effects Approach
• Potential omitted-variable bias in OLS model due to unobserved country-level
determinants of student performance that are correlated with teacher
cognitive skills (e.g., educational attitude)
• Exploit that both teacher skills and student performance are observed in two
distinct subjects
⇒ Allows to add student fixed effects
• Approach eliminates any non-subject-specific performance differences
between students (e.g., innate ability, family background, motivation)
• Also controls for any non-subject-specific differences across schools and
countries
Student Performance and Teacher Cognitive Skills (Student Fixed Effects)
Dependent variable: student performance difference: math – reading
Teacher skills: numeracy – literacy
(1)
(2)
(3)
0.105∗∗∗
0.117∗∗∗
0.106∗∗
(0.037)
(0.035)
(0.049)
Parent skills: numeracy – literacy
0.016
(0.035)
Instruction time: math – reading
Shortage teachers: math – reading
0.058∗∗
0.058∗∗
(0.026)
(0.026)
–0.012
–0.012
(0.012)
(0.012)
490,818
Students
490,818
490,818
Countries
31
31
31
0.01
0.02
0.02
Adj. R2
Exploiting Within-Country Variation in Cognitive Skills
• Potential bias in student FE model due to subject-specific differences across
countries (e.g., some countries particularly emphasize math skills)
• If there are subject-specific educational preferences, numeracy or literacy
skills in occupations other than teaching should also be higher
⇒ Control for cognitive skill level in other occupations (e.g., managers,
scientists and engineers, health professionals, business professionals, clerks,
sales workers, service workers)
• Results
Details
1. Cognitive skill level in only some occupations is significantly related to
student performance
2. Teacher cognitive skills always have strong impact on student
performance even when controlling for cognitive skill level of other
occupations
Determinants of Teacher Skills
Teacher Pay I
• Intuition: countries that pay teachers relatively better are able to recruit
teachers from higher up in the skill distribution and also are able to retain
teachers in their profession
• First, compute premium paid to teachers from standard Mincer regression:
lny = α0 + α1 G + α2 E + α3 E 2 + α4 A + δT + (1)
• lny = log gross hourly earnings
• G = gender
• E = potential work experience (age minus years of schooling minus 6)
• A = numeracy and literacy skills
• T = teacher indicator
• δ = teacher wage premium
IR
G L
R
KOC
LTR
ESU
P
IT
TU A
C R
A
D N
EU
PO
L
FI
FRN
JP A
CN
H
L
IS
SVR
BEN
G L
B
ESR
AUT
SG T
R P
U
NS
Z
AU L
N S
L
D D
N
C K
Z
N E
O
SVR
U K
S
SWA
E
Teacher Wage Premiums around the World
Estimated Teacher Premium
50%
40%
30%
20%
10%
0%
-10%
-20%
Teacher Pay II
• Then, relate teacher wage premium to teacher cognitive skills, controlling for
cognitive skills of all college graduates
• Accounts for international differences in overall country skill levels
• Exclude the ex-communist countries (Czech Republic, Estonia, Lithuania,
Poland, Russian Federation, Slovak Republic, Slovenia) and Turkey
• Occupational choices less driven by market forces but rather depended
on political attitudes
Teacher Wage Premiums and Teacher Cognitive Skills
Teacher skills in
Numeracy
Literacy
(1)
(2)
Teacher wage premium (/10)
0.113∗∗
0.097∗∗
(0.052)
(0.044)
Numeracy skills college graduates
0.943∗∗∗
(w/o teachers)
(0.112)
0.918∗∗∗
Literacy skills college graduates
(w/o teachers)
(0.070)
Countries
Adj. R2
∗
23
23
0.77
0.78
p<0.10, ∗∗ p<0.05, ∗∗∗ p<0.01
Conclusion
• Compute new, internationally comparable, measures of teacher numeracy and
literacy skills for 31 developed countries based on PIAAC data
• Compute proxies for cognitive skills of parents of PISA students
• Using various econometric approaches, we find that cross-country differences
in teacher cognitive skills are one important determinant of international
differences in student performance:
• 1 std. dev. increase in teacher numeracy (literacy) skills raises student
math (reading) performance by 14% (9%) of a std. dev.
• Raising numeracy skills of U.S. teachers to Finnish level would increase
student math performance by 33 PISA points (∼0.8 school years)
(Chile and Turkey: improvement by more than 1 school year)
• Results offer insights of teacher effectiveness that have previously been
unavailable
→ Smarter teachers produce smarter students
Suggestions for PIAAC Wave 2 to Improve Research on Teacher Effects
• Teacher-specific information
• Subject and grade level teachers are teaching
• Non-teacher-specific information
• Occupational tenure
• Mother’s and father’s occupation
• Keep question on college major
• Consider to oversample teachers
Thanks a lot for your attention!
[email protected]
Appendix
Are PIAAC Teacher Cognitive Skills Representative?
• In all countries, teachers have plausible skill levels
• Across all countries, teacher cognitive skills fall at 68th (70th ) percentile
of numeracy (literacy) skill distribution of all adults [ranging 53rd –78th
pct]
• Compare teacher skills from PIAAC with numeracy/literacy skills of teachers
in other nationally representative datasets with larger samples
• USA: combine NLSY 1979 + NLSY 1997
• NLYS birth cohorts partly cover birth cohorts in PIAAC
• Use math and language skills from four AFQT subtests
• Teacher cognitive skills in NLSY fall at 67th (64th ) percentile among
adult skill distribution in numeracy (literacy). PIAAC for USA: 70th
(71st ) pct
• Germany: adult cohort of National Educational Panel Study (NEPS)
• More than 9,300 adults, born 1944–1986, tested in 2010/2011
• Similar to PIAAC, tests aim at measuring skills in real-life situations
• NEPS: 67th (76th ) percentile. PIAAC for Germany: 72nd (74th ) pct
Back
Student Performance and Teacher Cognitive Skills
GRC
ISR
TUR
CHL
-2
-1
0
1
.8
.4
KORSGP
CAN
NZL
NLD AUS
POL
EST IRL BEL
NOR
DEU
FRAGBR
USA
DNK
SWE
ITA
CZE
ESP
SVN AUT
ISR GRC
LTU
SVK
RUS
0
FIN
-.4
-.4
ITA
RUS
JPN
NLD
BEL
DEU
NZL
AUS
DNK AUT
SVN
FRA
CZE
IRL
NOR
GBR SVK
SWE
ESP
USA
LTU
CAN
EST
TUR
JPNFIN
CHL
-.8
.4
0
POL
Reading performance student
.8
SGP
KOR
-.8
Math performance student
Without controls
2
-2
-1
Numeracy skills teacher
0
1
2
Literacy skills teacher
coef = .20897443, (robust) se = .03883733, t = 5.38
coef = .17820177, (robust) se = .02062038, t = 8.64
RUS
ISR
-1.5
-1
-.5
0
.5
Numeracy skills teacher
coef = .15106504, (robust) se = .06884994, t = 2.19
1
1.5
.8
.4
SGP
KOR
0
CAN
SVN IRL DEU FINFRA
NLD
ITA
AUS
ESPJPN
NZL
BEL
GBR
USA
TUR AUT
CZE
CHL
DNK
LTU
SVK NOR
GRC
SWE
DNK
-.4
.4
-.4
0
POL
EST
LTU
SVK
RUS
CAN
POL IRL
JPN
FRA
NZL
EST
ITA
AUS
TUR
DEU
BELUSA
ESP
NLD
GRC CHL
GBRSVN
ISR NOR
CZE
SWE
AUT
FIN
-.8
Reading performance student
.8
SGP
KOR
-.8
Math performance student
Controlling for adult cognitive skills
-1.5
-1
-.5
0
.5
Literacy skills teacher
coef = .2477728, (robust) se = .04786543, t = 5.18
1
1.5
Summary Statistics: Teacher Cognitive Skills
Pooled
Australia
Austria
Belgium
Canada
Chile
Czech R.
Denmark
Estonia
Finland
France
Numeracy
292
300
300
308
292
262
305
295
285
317
302
Literacy
295
312
292
303
307
263
300
288
294
322
296
Difference
-3
-12
8
5
-15
-1
5
7
-9
-5
6
Numeracy rank
68
71
69
68
67
81
73
56
60
73
80
Literacy rank
71
75
70
71
72
79
77
60
69
74
77
Observations
6,402
248
188
215
834
106
141
413
188
221
163
Germany
Greece
Ireland
Israel
Italy
Japan
Korea
Lithuania
Netherl.
New Zealand
Norway
Numeracy
308
282
295
270
273
311
287
285
304
297
302
Literacy
301
286
300
281
279
319
296
282
308
310
304
Difference
7
-5
-4
-12
-5
-8
-9
3
-4
-12
-2
Numeracy rank
72
74
75
57
67
70
72
66
63
64
65
Literacy rank
74
75
74
62
73
67
74
64
67
71
68
Observations
127
150
180
250
124
147
217
133
197
198
279
Poland
Russia
Singapore
Slovak R.
Slovenia
Spain
Sweden
Turkey
U.K.
U.S.
Numeracy
277
273
303
294
293
283
306
264
289
284
Literacy
293
283
300
290
288
290
307
261
299
301
Difference
-16
-10
3
4
5
-7
-1
3
-10
-17
Numeracy rank
64
53
72
66
70
75
62
80
65
70
Literacy rank
73
54
76
60
69
80
65
78
67
71
Observations
199
137
193
133
121
183
147
128
310
132
Summary Statistics: Parent Cognitive Skills
Pooled
Australia
Austria
Belgium
Canada
Chile
Czech R.
Denmark
Estonia
Finland
France
276
293
276
299
275
Numeracy
Mean
278
287
291
301
282
223
Mean
275
293
279
289
284
226
270
278
272
297
272
83,492
3,137
2,231
2,251
11,933
2,165
2,105
3,352
3,463
2,252
3,086
Germany
Greece
Ireland
Israel
Italy
Japan
Korea
Lithuania
Netherl.
New Zealand
Norway
276
277
295
284
297
307
281
271
293
288
290
Literacy
Observations
Numeracy
Mean
289
273
275
267
267
Mean
279
268
280
260
264
308
Literacy
Observations
2,293
2,128
2,371
1,882
1,789
2,103
3,361
2,364
2,276
2,504
2,228
Poland
Russia
Singapore
Slovak R.
Slovenia
Spain
Sweden
Turkey
U.K.
U.S.
Mean
264
271
261
281
265
295
240
281
267
Mean
267
277
253
275
261
266
290
237
285
277
1,793
1,074
2,119
2,442
2,435
2,614
1,864
2,319
3,578
1,980
Numeracy
268
Literacy
Observations
Summary Statistics: Country Characteristics
Pooled
Australia
Austria
Belgium
Canada
Chile
Czech R.
Denmark
Estonia
Finland
France
Expenditure per student
71
85
107
89
80
28
50
99
49
79
79
GDP per capita
35
41
43
40
40
19
28
42
23
39
36
School starting age
6.12
5
6
6
5
6
6
7
7
7
6
Instruction practice math
0.61
0.66
0.57
0.56
0.70
0.67
0.62
0.64
0.59
0.58
0.59
Instruction practice reading
0.50
0.53
0.41
0.43
0.56
0.53
0.44
0.57
0.50
0.37
0.52
Germany
Greece
Ireland
Israel
Italy
Japan
Korea
Lithuania
Netherl.
New Zealand
Norway
Expenditure per student
72
53
85
55
81
84
65
41
88
60
112
GDP per capita
40
28
44
30
35
34
30
21
45
32
61
School starting age
6
6
4
6
6
6
6
7
6
5
6
Instruction practice math
0.64
0.62
0.69
0.69
0.59
0.46
0.38
0.63
0.57
0.66
0.52
Instruction practice reading
0.44
0.49
0.51
0.40
0.49
0.44
0.34
0.58
0.37
0.53
0.37
Poland
Russia
Singapore
Slovak R.
Slovenia
Spain
Sweden
Turkey
U.K.
U.S.
Expenditure per student
49
13
78
43
85
78
89
16
91
111
GDP per capita
21
22
69
25
28
33
42
17
37
49
School starting age
7
7
7
6
6
6
7
7
5
6
Instruction practice math
0.60
0.69
0.70
0.54
0.56
0.64
0.51
0.59
0.73
0.72
Instruction practice reading
0.59
0.80
0.47
0.47
0.56
0.44
0.42
0.64
0.54
0.61
Complete List of Control Variables
Student characteristics
School characteristics
Age
Female
Migrant status (3 categories)
Other language spoken at home
School location (5 categories)
Private school
Number students per school
Content autonomy
Personnel autonomy
Budget autonomy
Shortage math/language teacher
Weekly hours math/language classes
Family background
Number of books at home (6 categories)
Education level (7 categories)
Occupation groups (4 groups)
Country-level measures
Educational expenditure per student
School starting age
Back
OLS Estimations: Results on All Other Covariates
Dependent variable: student performance
Student characteristics
Age
Female
First-generation migrant
Second-generation migrant
Other language at home
Family background
11-25 books
26-100 books
101-200 books
201-500 books
More than 500 books
ISCED 1
ISCED 2
ISCED 3B,C
ISCED 3A, 4
ISCED 5B
ISCED 5A, 6
Blue collar-high skilled
White collar-low skilled
White collar-high skilled
Math
∗∗∗
0.137
(0.018)
∗∗∗
–0.145
(0.011)
∗∗∗
–0.107
(0.038)
∗∗
–0.086
(0.035)
∗
–0.056
(0.029)
∗∗∗
0.186
(0.021)
∗∗∗
0.420
(0.033)
∗∗∗
0.588
(0.043)
∗∗∗
0.776
(0.049)
∗∗∗
0.775
(0.053)
∗∗∗
0.175
(0.042)
0.090
(0.065)
∗∗∗
0.254
(0.069)
∗∗∗
0.249
(0.062)
∗
0.169
(0.089)
∗∗∗
0.261
(0.085)
∗∗∗
0.119
(0.015)
∗∗∗
0.190
(0.016)
∗∗∗
0.403
(0.018)
Reading
∗∗∗
0.137
(0.012)
∗∗∗
0.358
(0.015)
∗∗
–0.103
(0.038)
–0.021
(0.034)
∗∗∗
–0.179
(0.031)
∗∗∗
0.226
(0.021)
∗∗∗
0.467
(0.034)
∗∗∗
0.647
(0.044)
∗∗∗
0.822
(0.053)
∗∗∗
0.801
(0.059)
∗∗∗
0.219
(0.042)
∗∗
0.137
(0.054)
∗∗∗
0.242
(0.060)
∗∗∗
0.270
(0.055)
∗∗∗
0.244
(0.074)
∗∗∗
0.330
(0.067)
∗∗∗
0.097
(0.018)
∗∗∗
0.184
(0.019)
∗∗∗
0.405
(0.020)
OLS Estimations: Results on All Other Covariates (continued)
Dependent variable: student performance
School characteristics
Small Town
Town
City
Large City
Private school
No. students per school (in 1000)
Content autonomy
Personnel autonomy
Budget autonomy
Shortage math teacher
Math
Reading
–0.008
(0.032)
0.014
(0.042)
0.014
(0.040)
∗
0.080
(0.045)
∗∗∗
0.140
(0.038)
∗∗∗
0.281
(0.062)
0.069
(0.051)
∗∗∗
–0.148
(0.048)
0.020
(0.039)
∗∗∗
–0.048
(0.012)
0.019
(0.028)
0.057
(0.035)
∗∗
0.079
(0.034)
∗∗∗
0.129
(0.043)
∗∗∗
0.159
(0.031)
∗∗∗
0.255
(0.052)
0.002
(0.032)
∗∗∗
–0.167
(0.031)
0.048
(0.036)
Shortage language teacher
Weekly hours math classes
∗∗
Weekly hours language classes
Country-level measures
Educational expenditure per student
School starting age
Students
Countries
Adj. R-squared
∗∗
–0.032
(0.013)
0.057
(0.027)
–0.001
(0.018)
–0.000
(0.001)
∗∗∗
0.139
(0.049)
0.000
(0.001)
∗
0.080
(0.041)
406,564
23
0.26
406,564
23
0.30
Student Performance and Teacher Cognitive Skills: Country-Level Skills
Student Math Performance
Teacher cognitive skills
Adult
Mean teacher
Parent
Adult
Mean teacher
Baseline
skills
skills
skills
Baseline
skills
skills
skills
(1)
(2)
(3)
(4)
(5)
(6)
(7)
∗∗∗
0.145
(0.032)
Parent cognitive skills
Student Reading Performance
Parent
∗∗∗
0.145
(0.048)
∗∗
0.125
(0.052)
∗∗
(0.031)
∗∗
0.044
0.042
(0.017)
Parent cognitive skills (country level)
∗∗∗
0.138
(0.016)
∗∗∗
0.092
(0.022)
∗∗∗
0.148
(0.040)
(0.045)
(8)
∗∗∗
0.093
(0.022)
0.015
0.013
(0.016)
(0.015)
0.030
–0.041
(0.036)
(0.035)
Adult cognitive skills (country level)
∗∗∗
0.150
0.051
–0.043
(0.040)
(0.041)
Student characteristics
X
X
X
X
X
X
X
X
Parent characteristics
X
X
X
X
X
X
X
X
School characteristics
X
X
X
X
X
X
X
X
Country characteristics
X
X
X
X
X
X
X
X
490,818
490,818
490,818
490,818
490,818
490,818
490,818
490,818
Students
Countries
Adj. R2
31
31
31
31
31
31
31
31
0.29
0.29
0.29
0.29
0.30
0.30
0.30
0.30
∗
Back
p<0.10, ∗∗ p<0.05, ∗∗∗ p<0.01
Student Performance and Teacher Cognitive Skills:
Including Additional Country Controls
Student Math Performance
(1)
Teacher skills
0.134
∗∗∗
(0.030)
Parent skills
∗∗
0.047
(0.017)
GDP p.c.
(2)
∗∗∗
0.138
(0.031)
∗∗
0.044
(0.017)
(3)
Student Reading Performance
(4)
∗∗∗
0.115
(0.028)
∗∗∗
0.108
(0.029)
∗∗∗
0.061
(0.016)
∗∗∗
0.052
(0.014)
∗∗∗
0.123
(0.027)
∗∗∗
0.056
(0.013)
∗∗∗
3.107
–15.861
(4.106)
Teacher wage
(5)
(3.883)
0.011
∗∗
∗∗∗
0.080
(0.024)
(7)
∗∗∗
0.084
(0.019)
0.022
0.015
(0.015)
(0.013)
(8)
(9)
∗∗∗
0.072
(0.023)
0.033
∗
(0.017)
(10)
∗∗∗
0.070
(0.021)
∗
0.032
(0.017)
∗∗∗
(0.018)
∗∗
0.031
(0.013)
–6.754
(3.156)
∗∗∗
∗∗∗
0.018
(0.006)
∗∗∗
0.068
∗∗
3.199
(2.651)
0.016
(0.005)
Perf. pay
(6)
0.018
(0.003)
(0.004)
–0.028
0.029
–0.049
0.003
(0.060)
(0.056)
(0.036)
(0.028)
∗∗∗
CEE
0.167
(0.056)
∗∗∗
∗∗
0.162
0.095
(0.047)
(0.036)
∗∗∗
0.092
(0.029)
Student char.
X
X
X
X
X
X
X
X
X
X
Parent char.
X
X
X
X
X
X
X
X
X
X
School char.
X
X
X
X
X
X
X
X
X
X
Country char.
X
X
X
X
X
X
X
X
X
X
Students
490,818
490,818
470,843
469,450
460,304
490,818
490,818
470,843
469,450
460,304
Countries
31
31
29
29
28
31
31
29
29
28
0.29
0.29
0.28
0.29
0.29
0.30
0.31
0.30
0.30
0.31
Adj. R2
∗
Back
p<0.10, ∗∗ p<0.05, ∗∗∗ p<0.01
Student Performance and Teacher Cognitive Skills
with Continental Fixed Effects and in Restricted Sample (OLS)
Student Math Performance
Teacher cognitive skills
Parent cognitive skills
Student Reading Performance
Continental
Europe
Continental
Europe
Baseline
fixed effects
only
Baseline
fixed effects
only
(1)
(2)
(3)
(4)
(5)
(6)
∗∗∗
0.145
∗∗∗
0.127
∗∗∗
0.104
∗∗∗
0.092
∗∗∗
0.088
(0.032)
(0.030)
(0.030)
(0.022)
(0.025)
∗∗
0.044
∗∗
0.024
∗∗
(0.029)
0.015
0.003
0.016
(0.014)
(0.016)
(0.012)
(0.015)
X
X
X
X
X
X
X
X
X
X
X
School characteristics
X
X
X
X
X
X
Country characteristics
X
X
X
X
X
X
Students
490,818
490,818
352,375
490,818
490,818
352,375
Countries
31
31
23
31
31
23
0.29
0.29
0.27
0.30
0.30
0.30
(0.017)
(0.012)
Student characteristics
X
Parent characteristics
Continental fixed effects
Adj. R2
X
∗
Back
0.038
∗∗
0.072
X
p<0.10, ∗∗ p<0.05, ∗∗∗ p<0.01
Instruction-Practice Indicators
• Students in PISA are asked about the frequency of activities their teachers do in language
classes (PISA 2009) or math classes (PISA 2012).
• Reading items (each measured on a 4-point scale ranging from “never or hardly ever” to “in
all lessons”) to construct the instruction-practice indicator:
•
•
•
•
•
•
•
asking students to explain the meaning of a text
asking questions that challenge students to get a better understanding of a text
giving students enough time to think about their answers
recommending books or author to read
encouraging students to express their opinion about a text
helping students relate the stories they read to their lives
showing students how the information in texts builds on what they already know
• Math items (each measured on a similar 4-point scale ranging from “never or rarely” to
“almost or almost always”):
• asking questions that make students reflect on the problem
• giving problems that require students to think for an extended time
• presenting problems in different contexts so that students know whether they have
understood the concepts
• helping students to learn from mistakes they have made
• asking students to explain how they have solved a problem
• presenting problems that require students to apply what they have learnt to new
contexts
Back
Exploiting Within-Country Variation in Cognitive Skills: Numeracy (OLS)
Dependent Variable: Student Math Performance
Occupation:
Managers
S&E
Health
Busin. I
Busin. II
Legal
(1)
(2)
(3)
(4)
0.004
0.053
(0.025)
(0.038)
(0.033)
(0.030)
(0.027)
(0.038)
0.28
0.28
0.28
0.29
0.29
0.28
(1)
(2)
(3)
(4)
–0.009
0.024
Clerk
Service
Sales
Panel A: Without Teacher Skills
Occupation skills
Adj. R2
∗∗∗
0.096
(5)
∗∗∗
0.112
(6)
∗∗∗
(7)
∗
0.156
0.070
(8)
(9)
0.048
0.068
(0.031)
(0.039)
(0.043)
0.29
0.28
0.28
∗∗∗
0.133
Panel B: With Teacher Skills
Occupation skills
–0.037
(0.062)
Teacher skills
∗∗
0.172
∗∗
–0.105
(0.049)
∗∗∗
0.199
(0.038)
∗∗∗
0.148
(0.042)
∗∗
0.126
(5)
∗∗
0.093
(0.044)
∗∗
0.091
(6)
–0.009
(0.039)
∗∗∗
0.149
(7)
∗
0.074
(0.043)
∗∗
0.112
(8)
(9)
–0.018
0.005
(0.046)
∗∗∗
0.152
(0.048)
∗∗∗
0.143
(0.069)
(0.043)
(0.039)
(0.049)
(0.042)
(0.043)
(0.041)
(0.043)
(0.043)
Students
490,818
490,818
490,818
490,818
490,818
490,818
490,818
490,818
490,818
Countries
31
31
31
31
31
31
31
31
31
0.29
0.29
0.29
0.29
0.29
0.29
0.29
0.29
0.29
Adj. R2
Exploiting Within-Country Variation in Cognitive Skills: Literacy (OLS)
Dependent variable: Student Reading Performance
Occupation:
Managers
S&E
Health
Busin. I
Busin. II
(1)
(2)
(3)
(4)
0.034
0.013
0.025
(0.026)
(0.025)
(0.033)
(0.030)
0.30
0.30
0.30
0.30
(1)
(2)
(3)
(4)
0.020
–0.042
∗
–0.079
–0.065
0.020
(0.053)
(0.032)
(0.040)
(0.042)
(0.039)
Legal
Clerk
Service
Sales
Panel A: Without Teacher Skills
Occupation skills
Adj. R2
∗∗∗
0.081
(5)
(6)
(7)
(8)
(9)
0.029
0.021
–0.003
–0.003
(0.025)
(0.029)
(0.028)
(0.029)
(0.032)
0.30
0.30
0.30
0.30
0.30
(6)
(7)
(8)
(9)
–0.040
–0.044
∗∗
–0.070
(0.035)
(0.035)
(0.034)
∗∗∗
0.082
Panel B: With Teacher Skills
Occupation skills
Teacher skills
0.075
∗∗∗
0.118
∗∗∗
0.141
∗∗∗
0.147
(5)
∗∗
0.078
∗∗∗
0.118
∗∗∗
0.113
∗∗∗
0.129
∗∗
–0.066
(0.029)
∗∗∗
0.125
(0.046)
(0.030)
(0.035)
(0.044)
(0.034)
(0.031)
(0.032)
(0.029)
(0.028)
Students
490,818
490,818
490,818
490,818
490,818
490,818
490,818
490,818
490,818
Countries
31
31
31
31
31
31
31
31
31
0.30
0.30
0.30
0.30
0.30
0.30
0.30
0.30
0.30
Adj. R2
Back
Additional Statistics
• Teacher descriptives
•
•
•
•
72% of teachers
89% of teachers
75% of teachers
Average teacher
are female (ranging from 53% to 90%)
are college graduates (ranging from 67% to 98%)
are public sector employees
age: 41.8 years
• Teacher cognitive skills
• 1 SD = 14.2 (13.8) points in numeracy (literacy)
• international individual-level standard deviation in PIAAC = 55 points
(numeracy)/50 points (literacy)
• return to skills in pooled sample: almost 4% higher earnings for raise of
10 PIAAC points
• Skill correlations
• teacher vs. adult skills (country level): 0.77 (0.86) numeracy (literacy)
• teacher vs. parent cognitive skills (student level): 0.40 (0.43) numeracy
(literacy)
• teacher numeracy vs. teacher literacy skills = 0.85
• 4-digit and 2-digit ISCO codes: 0.97 numeracy, 0.95 literacy
• parent numeracy vs. literacy skills (at student level) = 0.94