Can Public Policies Lower Religiosity? Evidence from School Choice in France, 1878–1902

1Argument

Between 1878 and 1902 the French Third Republic set out to move children from Catholic primary schools into secular state schools. It used two kinds of policy. It spent heavily on secular schools, and central state spending on education rose fourteenfold between 1875 and 1902. It also passed laws. The laws of 1881 and 1882 made state primary schooling free, secular, and mandatory to age 13. The law on associations of 1 July 1901 in effect barred monks and nuns, who supplied most Catholic teachers, from teaching.

The paper uses the share of pupils in Catholic primary schools as a measure of religiosity, across the 87 départements of France in 13 years with complete data. The added spending had no robust effect on that share. The laws of 1881 and 1882 lowered it a little. The 1901 law lowered it sharply, especially for girls. Parents who chose Catholic schools were not won over by cheaper and better-equipped secular schools. They moved their children when the law removed the teachers.

Catholic schooling held on most strongly where the Revolution's violence against the Church had been heaviest. The paper traces its resilience to a political divide dating from 1789.

2The laws

The Republic's school policy came in two waves. The first, from 1878 to 1886, paid for secular schools and teachers and made state schooling free and compulsory. The second, from 1901 to 1905, went after the religious orders that staffed Catholic schools.

3Spending and enrolment

Public spending on primary education nearly tripled between 1878 and 1902 in the paper's data, and the central state took over most of it. Spending by the départements fell to almost nothing.

The share of pupils in Catholic schools fell slowly through the 1880s and 1890s. Girls were far more likely than boys to attend a Catholic school. In 1878 the average département sent 19 percent of its boys and 55 percent of its girls to Catholic schools. In 1896 the figures were 15 and 42 percent. Then the 1901 law took effect, and by 1902 they were 12 and 27 percent.

4Catholic schooling by département

The averages hide large differences between départements. In the most Catholic départements close to 60 percent of boys and 92 percent of girls attended Catholic schools in 1878. The regional pattern held through 1902.

5What moved enrolment

The paper regresses the share of boys or girls in Catholic schools on public spending, a marker for the years after the 1881 law, and a marker for the years after the 1901 law. Every specification compares each département with itself over time. Some add controls for fertility, literacy, urbanization, and the structure of the workforce. Some absorb common time patterns with a time trend and some with year effects. Choose a specification to see its estimates.

Spending in the average département rose by about 0.86 log points between 1878 and 1902, a factor of about 2.4. Multiplying that rise by the spending coefficient gives the change in the Catholic share it predicts. Across the eight specifications the answer is a fall of 0.9 to 1.7 percentage points, and the estimate is significant at the 10 percent level only for boys with a time trend. Over the same years the 1901 law lowered the share of girls in Catholic schools by about 5.5 percentage points against the time trend, from an average share of 44 percent. For boys the estimate is 0.7 to 1.0 points.

6Where Catholic schooling held

The paper asks where spending did least to move parents. It interacts spending with characteristics of each département that predate the Republic, one at a time. A positive interaction means spending lowered the Catholic share less where the characteristic was larger. Choose a characteristic to see how it lines up with the average Catholic share, 1878 to 1902, and what the paper estimates.

The strongest pattern runs through the Revolution. Spending did less to move girls out of Catholic schools in the départements with the most death sentences in the Terror of 1793 and 1794, and in those where the Terror fell most on the clergy. The death sentences were concentrated in the three largest cities and in Loire-Inférieure, Maine-et-Loire, and Vendée, where the revolutionary state carried out mass killings. Spending also did less to move girls in rugged départements, and less to move boys where primary school attendance had been high in 1837. The share of priests who refused the 1791 oath to the civil constitution of the clergy, a measure of religiosity before the violence, does not predict the effect of spending.

7What it says about policy

The Republic competed with the Catholic schools on price and quality. Secular schooling became free and there were more state schools and teachers. Parents who preferred Catholic schools mostly stayed. The share fell sharply only when a law removed the supply of Catholic teachers. In the paper's words, Catholic parents were not won over by free and mandatory schooling but were coerced into enrolling their children in secular schools.

The paper reads this as a case in which the state failed to crowd out a private provider by competing on price. Making secular education cheaper by building schools and hiring teachers was not enough to break the links between religion and politics that dated from the Revolution's assault on the Church a century earlier. Only a restriction on who could teach moved large numbers of children.

8Limits

Enrolment in Catholic schools measures religiosity indirectly. The paper's online appendix shows it correlates with other historical measures of religiosity, but school choice also reflects tuition, distance, and the quality of the local schools.

Spending and enrolment are observed together in only 13 years, 1878 to 1882, 1886 and 1887, 1896 and 1897, and 1899 to 1902. The data say nothing about schooling after the 1904 law or the 1905 separation of church and state.

The laws applied to every département in the same year, so their estimated effects rest on timing alone. They cannot be separated from anything else that changed nationally in 1881 or in 1901 and 1902.

The channels in Section 6 are correlations across départements. They describe where spending worked least and do not establish why.