### Table 4. OLS estimations

"... In PAGE 18: ... If we assume that the error term is independent of X, this model can be estimated by ordinary least squares (OLS). Table4 shows the results of this estimation, which could be directly comparable to previous cross-sectional evidence in vertical integration. 35 Subcontracting tends to decrease as hold-up problems grow and to increase with the number of provinces in which the firm operates, its diversity of output, and its degree of specialization in design and technical management.... In PAGE 18: ... Uncertainty does not seem to have significant effects on subcontracting. Finally, although all the regressions presented in Table4 contain year dummies for the reporting year for each observation, they are not shown because none of them is significant. This estimation, however, does not control for the unobservable heterogeneity, which is the main advantage of panel data.... ..."

### Table 5: OLS Estimates of

1997

### Table 10. OLS Estimates of Teacher Absence

"... In PAGE 15: ... Since OLS coefficients are easier to interpret (especially relative to an ordered probit). Table10 presents the results from OLS regressions using the visit level and teacher level dependent variables. Columns 1 and 3 use state per capita income on the right hand side, while columns 2 and 4 include state fixed effects.... In PAGE 15: ... The regressions also control for the visit effect and for the time of day effect. The discussion of the regression results reported in Table10 is organized according to the hypotheses laid out in section 2. 4.... In PAGE 17: ... Table10 (continued. OLS Estimates of Teacher Absence Dependent variable (visit level observation): 0 = present, 100 = absent Dependent variable (teacher level observation): average absence (over 3 visits) multiplied by 100 [1] [2] [3] [4] No state fixed effects With state fixed effects No state fixed effects With state fixed effects Mean fathers education of 4th grade children (1- 7 scale) -0.... In PAGE 18: ... We therefore construct an infrastructure index from 1 to 5 with each of the aforementioned facilities being worth 1 point. Under this specification (going back to Table10 ), each point on the index is associated with a 1.... ..."

### Table 10. OLS Estimates of Teacher Absence

"... In PAGE 15: ... Since OLS coefficients are easier to interpret (especially relative to an ordered probit). Table10 presents the results from OLS regressions using the visit level and teacher level dependent variables. Columns 1 and 3 use state per capita income on the right hand side, while columns 2 and 4 include state fixed effects.... In PAGE 15: ... The regressions also control for the visit effect and for the time of day effect. The discussion of the regression results reported in Table10 is organized according to the hypotheses laid out in section 2. 4.... In PAGE 17: ... Table10 (continued. OLS Estimates of Teacher Absence Dependent variable (visit level observation): 0 = present, 100 = absent Dependent variable (teacher level observation): average absence (over 3 visits) multiplied by 100 [1] [2] [3] [4] No state fixed effects With state fixed effects No state fixed effects With state fixed effects Mean fathers education of 4th grade children (1- 7 scale) -0.... In PAGE 18: ... We therefore construct an infrastructure index from 1 to 5 with each of the aforementioned facilities being worth 1 point. Under this specification (going back to Table10 ), each point on the index is associated with a 1.... ..."

### Table 10 (continued. OLS Estimates of Teacher Absence

"... In PAGE 15: ... Since OLS coefficients are easier to interpret (especially relative to an ordered probit). Table10 presents the results from OLS regressions using the visit level and teacher level dependent variables. Columns 1 and 3 use state per capita income on the right hand side, while columns 2 and 4 include state fixed effects.... In PAGE 15: ... The regressions also control for the visit effect and for the time of day effect. The discussion of the regression results reported in Table10 is organized according to the hypotheses laid out in section 2. 4.... In PAGE 16: ...significantly more likely to be absent having an absence rate of 4-5% more than regular teachers (even after controlling for age and education). Table10 . OLS Estimates of Teacher Absence Dependent variable (visit level observation): 0 = present, 100 = absent Dependent variable (teacher level observation): average absence (over 3 visits) multiplied by 100 [1] [2] [3] [4] No state fixed effects With state fixed effects No state fixed effects With state fixed effects Gender (1 = Male) 1.... In PAGE 18: ... We therefore construct an infrastructure index from 1 to 5 with each of the aforementioned facilities being worth 1 point. Under this specification (going back to Table10 ), each point on the index is associated with a 1.... ..."

### Table 10 (continued. OLS Estimates of Teacher Absence

"... In PAGE 15: ... Since OLS coefficients are easier to interpret (especially relative to an ordered probit). Table10 presents the results from OLS regressions using the visit level and teacher level dependent variables. Columns 1 and 3 use state per capita income on the right hand side, while columns 2 and 4 include state fixed effects.... In PAGE 15: ... The regressions also control for the visit effect and for the time of day effect. The discussion of the regression results reported in Table10 is organized according to the hypotheses laid out in section 2. 4.... In PAGE 16: ...significantly more likely to be absent having an absence rate of 4-5% more than regular teachers (even after controlling for age and education). Table10 . OLS Estimates of Teacher Absence Dependent variable (visit level observation): 0 = present, 100 = absent Dependent variable (teacher level observation): average absence (over 3 visits) multiplied by 100 [1] [2] [3] [4] No state fixed effects With state fixed effects No state fixed effects With state fixed effects Gender (1 = Male) 1.... In PAGE 18: ... We therefore construct an infrastructure index from 1 to 5 with each of the aforementioned facilities being worth 1 point. Under this specification (going back to Table10 ), each point on the index is associated with a 1.... ..."

### Table 3: OLS Estimates of the Effect of CAI Intensity

"... In PAGE 11: ...omponent, , , is specific to pupils. The coefficient, quot;, is the parameter of primary interest. is Fourth graders in schools where teachers report using more CAI generally have higher Math scores, but there is little evidence of an association between CAI and Hebrew scores in either grade. This can be seen in Table3 , which reports OLS estimates of the relationship between CAI intensity and test scores for applicants, for applicants with test score data, and for a sample of pupils in large towns. This last sample is used to control for town fixed effects, and includes any pupil (whether or not their school applied for Tomorrow funds) living in a town with at least two schools.... In PAGE 12: ...8. Except for the Hebrew scores of fourth graders, Table3 shows a pattern of declining effects as the models included larger sets of schools, i.e.... In PAGE 27: ...34 (1.37) N 2620 2145 2620 2145 2593 2135 2593 2135 Other included controls Pre-existing computers X X X X X X X X Basic controls X X X X X X X X 1991 test scores X X X X Notes: Basic controls and lagged test scores: as in Table3 . Standard errors are reported in parentheses.... In PAGE 28: ... The computer use-intensity ranking is equal to 0 if teacher never uses computer, =1 if sometimes, =2 if often, =3 if always. Basic controls: and lagged test scores as in Table3 . The sample in columns 3 and 6 is limited to pupils in schools that received T-98 funding, including those that received funding after the June 1996 test date (as of 1998).... In PAGE 29: ...robability is a nonparametrically estimated function of the normalized town rank for funding. Estimates use the bandwidth shown. All models control for a quadratic function of the normalized rank and for the number of applicants in the town. Basic controls: as in Table3 . The samples in columns 3 and 6 are limited to pupils in schools with normalized town ranking for Tomorrow-98 funding above .... ..."

### Table 2: OLS Estimates of Equation 1

1997

"... In PAGE 6: ...Table2 shows the results of estimating equation n281n29 from 1981 through 1995 using St. Louis and Board measures of home base and total base.... ..."

### Table 3. OLS Estimates of quadratic model coefficients

1994

Cited by 10