increases by 0.1 log-points (approximately 10 percent) student achievement declines by 0.15
“core” country standard deviations.
|
Table 4. The effect of government consumption on student achievement | ||||||||
|
(1) |
(2) |
(3) |
(4) |
(5) |
(6) |
(7) |
(8) | |
|
Gov. consumption spending, |
0.107 |
-1.462** |
-0.650 |
-1.021* |
-1.596** |
-1.100** |
-1.060 |
-1.086 |
|
percent of GDP (log) |
(0.414) |
(0.378) |
(0.509) |
(0.486) |
(0.558) |
(0.377) |
(0.945) |
(1.551) |
|
GDP per capita (log) |
- |
1.502** |
1.234* |
1.123* |
2.056** |
0.483 |
1.451 |
3.735* |
|
(0.213) |
(0.485) |
(0.475) |
(0.748) |
(0.292) |
(0.698) |
(1.473) | ||
|
Percentage secondary school |
- |
0.763** |
0.510 |
0.736* |
2.238** |
1.268** |
0.424 |
1.454 |
|
attained among adults (log) |
(0.292) |
(0.409) |
(0.371) |
(0.836) |
(0.271) |
(0.639) |
(1.052) | |
|
Population size (log) |
- |
0.040 |
2.279* |
1.515 |
-8.492 |
-0.048 |
5.221* |
-11.04 |
|
(0.77) |
(1.046) |
(0.978) |
(6.758) |
(0.62) |
(2.241) |
(8.728) | ||
|
Trend |
- |
- |
- |
-0.041** |
- |
- |
- |
- |
|
(0.014) | ||||||||
|
Country fixed effects |
No |
No |
Yes |
Yes |
Yes |
No |
Yes |
Yes |
|
Time fixed effects |
No |
Yes |
Yes |
No |
Yes |
Yes |
Yes |
Yes |
|
County specific trends |
No |
No |
No |
No |
Yes |
No |
No |
Yes |
|
Observations |
232 |
208 |
197 |
197 |
187 |
128 |
128 |
124 |
|
No of countries |
72 |
59 |
48 |
48 |
43 |
28 |
28 |
26 |
|
Sample |
All |
All |
All |
All |
All |
OECD |
OECD |
OECD |
|
R2 |
0.0003 |
0.455 |
0.943 |
0.937 |
0.982 |
0.301 |
0.851 |
0.932 |
|
R2 (within) |
- |
- |
0.222 |
0.133 |
0.740 |
- |
0.298 |
0.679 |
|
Note. Absolute standard errors in parentheses, +, * and ** denote significance at 10, 5 and 1 percent level, | ||||||||
|
respectively. | ||||||||
Columns (3) to (5) of Table 4 present models with country fixed effects that mitigate a
potential omitted variable problem.16 In particular, the fixed effects approximate various
omitted factors such as organization of schools, curriculum, school autonomy and
centralization of exams. In the specific context of Table 4, they also capture that student
achievement is highest in rich and well-educated countries. However, including country fixed
effects in addition to year effects in column (3) does not change the point estimates of GDP
and adult education attainment much compared to column (2), although the standard errors are
twice as large as in the model with only time fixed effects (column (2)). On the other hand,
the effect of government consumption is reduced and becomes insignificant.
16 The number of observations is reduced in the fixed effects models because only countries with at least two
observations can contribute to the identification.
16