Does trade shape educational decisions? The role of initial schooling

Asia’s trade share includes South Asia, which reduces the overall average.

Primary school enrolment rates in the MENA region are 93 percent (85 percent for female and 101 percent for male); secondary school enrolment rates are 59 percent (53 percent for female and 65 percent for male); tertiary enrolment rates are 19 percent (18 percent for female and 21 percent for male). Primary school enrolment rates in LAC are 113 percent (112 percent for female and 115 percent for male); secondary school enrolment rates are 75 percent (77 percent for female and 73 percent for male); tertiary enrolment rates are 25 percent (26 percent for female and 23 percent for male). Primary school enrolment rates in SSA are 81 percent (75 percent for female and 87 percent for male); secondary school enrolment rates are 27 percent (24 percent for female and 30 percent for male); tertiary enrolment rates are 4 percent (3 percent for female and 5 percent for male). Primary school enrolment rates for Transition economies are 98 percent (98 percent for female and 99 percent for male); secondary school enrolment rates are 90 percent (90 percent for female and 90 percent for male); tertiary enrolment rates are 35 percent (40 percent for female and 40 percent for male). Primary school enrolment rates in Asia are 107 percent (104 percent for female and 111 percent for male); secondary school enrolment rates are 56 percent (55 percent for female and 60 percent for male); tertiary enrolment rates are 14 percent (15 percent for female and 15 percent for male).

The idea here is consistent with Romer’s (1990) Noble Prize idea of “Endogenous Technological Change” whereby the starting point for countries to generate sustainable economic growth turns on the stock of human capital, which feeds into a virtuous circle of accumulated knowledge and stronger economic prosperity.

Heckman (2006) provides a cogent exposition of why initial knowledge incentivised future learning: (1) early learning confers value on acquired skills, which leads to self-reinforcing motivation to learn more; (2) early mastery of a range of cognitive, social, and emotional competencies makes learning at later ages more efficient and therefore easier and more likely to continue.

There is theoretically a competing reason why the average U.K’s household may not likely self-select into the labour market: the U.K’s household also has a higher baseline income, so they can absorb the cost associated with more schooling. The placebo exercise in Table 13 of the Appendix supports the role of initial education.

Countries that received large remittance inflows, remittances exert strong influence on the impact that trade has on government consumption (Williams 2021a, b).

The Barro and Lee (2013) education data ended 2015. I use interpolated data up to 2021.

I used the 2SLS IV estimation rather than the GMM for several reasons. First, the superiority of the 2SLS IV method is that it uses external instruments to isolate the exogenous variation of the endogenous variable, while the GMM uses internal instruments (lags). In other words, the GMM uses lags (internal) of the endogenous variable as instruments. Using internal instruments are problematic because, by construction, lags of the endogenous variables are correlated with the endogenous variables themselves and thus lags of the endogenous variables will also be endogenous to the outcome variable. Second, there are three sources of endogeneity—reverse causality, measurement errors, and omitted variables—and the 2SLS IV estimation resolves all three, while the GMM estimation addresses only reverse causality using internal instruments. Third, the GMM estimation can suffer from instrument proliferation and thus bias the estimates. For these reasons, I used the 2SLS IV estimation to address endogeneity concerns.

Stock and Yogo (2005) recommended F-statistic above 10, below which instruments are considered weak.

I thank an anonymous referee for suggesting this exercise.

Men have stronger preference for more children than women (Bankole 1995; Bankole and Singh 1998; Westoff 2010; Field et al. 2016; Doepke and Tertilt 2018).

Doepke et al. (2022) suggest three additional reasons why one should expect women to combine a career with large families: “greater contributions from fathers in providing childcare; social norms in favour of working mothers; and flexible labour markets.”.

See also Williams (2021a, b) who show that national income mediates the link between trade and government size.