The effects of foreign product demand-labor transfer nexus on human capital investment in China

Over the past two decades, China has experienced substantial labor transfer (LT) from the agricultural to the non-agricultural sector, coinciding with rapid growth in foreign product demand (FPD). The study investigates whether and how increases in FPD drive LT and, in turn, affect individuals’ human capital investment decisions. The rationale is twofold: non-agricultural jobs typically require higher skills than agricultural work, and wages are generally higher outside agriculture, changing both the costs and expected returns to human capital. The research formulates three hypotheses: H1, that higher FPD promotes LT from agriculture to non-agriculture; H2, that LT positively influences individuals’ human capital investment; and H3, that LT increases income and thereby raises children’s human capital investment. The study contributes by focusing specifically on the FPD-induced LT mechanism and examining gender and intergenerational implications using large-scale micro and macro data across 31 Chinese provinces and municipalities from 2001–2021.

Prior studies link trade openness to human capital through three channels: skill premium (changes in demand for different skill levels), income effects (ability to pay vs. opportunity costs), and competitive pressures (firms upgrading and demanding higher skills). Evidence is mixed: openness can raise skill premia and educational attainment (e.g., in the U.S.), but can also raise demand for low-skilled labor and reduce schooling (e.g., Mexico). Income growth can both enable education spending and raise opportunity costs, potentially hindering human capital investment. Existing work provides limited evidence on the specific role of LT induced by FPD on human capital. This study departs from the trade openness focus to analyze FPD directly, arguing that rising FPD in China expands non-agricultural demand, draws idle agricultural labor into non-agricultural sectors, requires higher skills, and raises incomes, thereby affecting both adult and children’s human capital investments, including gender equality and intergenerational differences.

Theoretical framework: An FPD–LT model is developed with two sectors (agricultural A and non-agricultural NA). Output depends on labor only: QA = L_A^α and QNA = L_NA^αNA with L_A + L_NA = L. Wages derive from profit maximization. In a closed economy, equilibrium occurs when real wages equalize across sectors. In an open economy, comparative advantage shifts relative prices; since China primarily exports non-agricultural goods, the relative price of non-agricultural goods rises, shifting equilibrium toward higher L_NA and lower L_A. Hypothesis H1 posits that higher FPD increases LT to non-agriculture. Cost–benefit analysis of human capital investment is specified via discounted benefit R(t) and cost C(t) functions incorporating wages for high-skilled (WNA) and low-skilled (WA), education duration (E), and parameters capturing skill level and education cost. Solving R(t)=C(t) yields an expression linking optimal skill level to wage and labor ratios; greater LT (higher L_NA/L_A) increases the incentive to invest in human capital (supporting H2). A simple model of children’s human capital investment (CHCI) assumes a fraction p of income is spent on children’s education; with income increasing with skill acquisition and LT, CHCI rises (H3). Empirical strategy: The main regression for individual human capital investment (education years) is humancapital_ict = β0 + β1 LT_t−1 + province and year fixed effects + error, estimated by OLS and then by 2SLS with an instrumental variable (IV) to address endogeneity. The one-period lag of FPD (measured as ln(exports/GDP)) serves as the IV for LT, justified because FPD affects labor demand but not directly the supply of labor or individuals’ education length. Heteroscedasticity is tested via Breusch–Pagan and addressed using weighted least squares (WLS). Additional models assess effects on children’s education expenditure (log) and on income: Ineduexp_ict = a2 + β2 LT_t−1 + FE + error; income_ict = a1 + β1 LT_t−1 + FE + error, using the same IV strategy where appropriate. Heterogeneity analyses examine differences by gender and parental education (junior college and above vs. lower). Data: Macro provincial panel data (31 provinces/municipalities) from 2001–2021 (National Bureau of Statistics; China Urban Statistics Yearbook 2002–2022) measure FPD and LT. LT is defined as the ratio of the change in total employment minus the change in agricultural employment to total employment; positive values indicate transfer into non-agriculture. Micro data on individuals (income, education years, child education expenditure, gender, education background) come from CHIPS 2002, 2007, 2008 and CFPS 2010, 2012, 2013, 2014, 2016, 2018, 2020, yielding 72,915 observations after preprocessing and 1% winsorization of continuous variables. Fixed effects for year and province are included, and robust standard errors reported. Robustness checks employ provincial-level analyses of FPD→LT and LT→average education metrics.

- FPD increases labor transfer to non-agriculture (H1): At the provincial level, a 1% increase in FPD is associated with a 0.025% increase in the overall employment rate and a 0.032% increase in the non-agricultural employment rate (both 1% significance). - LT increases individual human capital investment (H2): Micro-level regressions show that a 1% increase in LT is associated with higher education years—OLS: +4.720% (SE 0.506, 1% sig), WLS: +2.271% (SE 0.712, 1% sig), and 2SLS-IV: +5.902% (SE 3.293, 10% sig). - Gender heterogeneity: A 1% increase in non-agricultural employment rate raises male education years by 3.890% and female education years by 5.453% (both 1% sig), indicating greater gains for females and movement toward gender equality. - LT raises incomes: OLS and WLS show positive effects; 2SLS-IV indicates that a 1% increase in LT increases income by 21.264% (SE 5.901, 1% sig). - Higher income increases children’s education expenditure (H3): For every 1% increase in income, children’s education expenditure rises by 5.313% (1% sig). By period: 2002–2008, +5.415% (1% sig); 2010–2020, +6.980% (1% sig), indicating growing emphasis on children’s human capital investment over time. - Parental education heterogeneity: For each 1% increase in LT, children’s education expenditures increase by 5.698% for highly educated parents (junior college and above) and by 3.310% for lower-educated parents (both 1% sig), suggesting potential widening intergenerational inequality. - Provincial robustness: Non-agricultural employment rate is positively related to average educational levels (e.g., +0.958% in high school share, +0.684% in junior college+ share, and +0.097% in average education years per 1% increase), while overall employment rate is negatively related to education measures, consistent with opportunity cost mechanisms.

The findings support the theoretical mechanism whereby rising FPD shifts labor demand toward non-agriculture, inducing labor transfer that increases both the returns to and the need for human capital investment. The positive effect of LT on education years indicates that higher skill requirements in non-agricultural sectors spur educational investment. The stronger response among females suggests LT contributes to narrowing gender gaps in human capital. Incomes increase with LT, and higher incomes translate into higher spending on children’s education, confirming the income channel of intergenerational human capital investment. Provincial analyses show that while higher overall employment can reduce education via higher opportunity costs, increases specifically in non-agricultural employment correlate positively with educational attainment, underscoring the sectoral composition’s importance. Overall, the results demonstrate that FPD-induced structural change affects education decisions directly and through income, with implications for gender equality and potential intergenerational inequality.

This study links foreign product demand, labor transfer, and human capital investment in China using theoretical modeling and extensive micro- and macro-data. Key conclusions: (1) FPD significantly promotes labor transfer from agriculture to non-agriculture; (2) LT increases individuals’ education years, with greater gains for females, fostering gender equality in education; (3) LT raises incomes, which increases children’s education expenditures; and (4) higher-educated parents invest more in their children’s education than lower-educated ones, potentially exacerbating intergenerational inequality. Policy recommendations include enhancing comparative advantages—especially in technology-intensive exports—to stimulate high-skill demand; improving social welfare (healthcare, children’s education) for workers transitioning from agriculture; and paying special attention to human capital investment in agricultural regions to avoid widening gaps. Future research should differentiate types of FPD and examine transfers across non-agricultural sub-sectors with varying skill requirements, particularly in the context of advancing artificial intelligence.

The study does not distinguish among different types of FPD (e.g., technology-intensive vs. labor-intensive), which may have heterogeneous effects on skill demand and education. It also aggregates non-agricultural sectors, overlooking variation across sub-sectors with differing skill requirements; transfers into lower-skill non-agricultural sub-sectors may not promote human capital investment. Future work should disaggregate FPD composition and sectoral destinations of LT, and assess implications under increasing AI adoption.