Education Policy and Long-Run Productivity

Why Education Policy Matters for Productivity

Education policy can influence an economy through more than one channel. Schooling may improve a worker’s ability to perform existing tasks; vocational training may help workers use particular technologies; and higher education may build skills used in research and innovation. Each channel can raise productivity, but the implications for long-run growth depend on how productivity changes and whether the change continues.

A key distinction is between a higher level of output and a higher rate of growth. If a policy raises workers’ skills once, output may move to a permanently higher path while its growth rate eventually returns to its previous value. If education permanently changes the rate at which human capital or technology accumulates, it may raise the economy’s sustained growth rate in a model that allows that mechanism.

Definition: Human capital is the productive knowledge, skills, and abilities embodied in people. An education policy raises human capital when it improves these productive capabilities. A productivity-level effect changes output per unit of input at a point in time; a growth-rate effect changes the ongoing rate at which output or productivity increases.

This tutorial focuses on those policy channels and the level-versus-growth distinction. The conclusions depend on the model’s assumptions: education can affect measured output only if the acquired skills are productive and are used, and a growth-rate effect requires a continuing accumulation or innovation mechanism. This treatment is aligned with the SOA VEE Economics (Validation by Educational Experience) syllabus.

Education, Human Capital, and Output

A useful production framework is \(Y=A K^\alpha (hL)^{1-\alpha}\), where \(Y\) is output, \(A\) is technology, \(K\) is physical capital, \(L\) is the number of workers, and \(h\) is human capital per worker. The parameter \(\alpha\), between zero and one, is the output elasticity of physical capital. The term \(hL\) represents human-capital-adjusted labor. In this framework, higher \(h\) raises effective labor, while a change in \(A\) raises productivity beyond the measured quantities of capital and labor.

The framework is a simplification. It treats workers as having a common human-capital level; real economies have different skills, occupations, and complementarities between workers and equipment. It also does not say that every additional year of education has the same effect. The size and direction of a policy effect depend on the skills taught, who receives them, the quality of instruction, and whether workplaces can use those skills.

y = A k^\alpha h^{1-\alpha}

Here \(y=Y/L\) is output per worker and \(k=K/L\) is physical capital per worker. Taking growth rates gives a useful accounting relationship. Holding the production function’s shares constant, the growth rate of output per worker depends on the growth rates of technology, capital per worker, and human capital per worker.

g_y = g_A + \alpha g_k + (1-\alpha)g_h

This relationship does not by itself establish what causes \(g_A\), \(g_k\), or \(g_h\). It shows how changes in those inputs map to output growth under the stated production function. If a policy changes \(h\) once and then \(h\) stops growing faster, the policy contributes to a level difference, not a permanently higher \(g_h\).

Worked Example: Converting Additional Schooling Into a Productivity Level

Method: human-capital index. Consider a hypothetical model in which an individual’s human-capital index is \(h=e^{rs}\), where \(s\) is completed years of schooling and \(r=0.06\) is a constant return parameter per year. Assume the policy leads a worker to complete three additional years of schooling, the skills are fully used at work, and the model’s relationship applies to this worker. Calculate the productivity ratio after the additional schooling. The policy does not change the worker’s subsequent rate of skill accumulation.

Let \(h_0\) be the index before the additional schooling and \(h_1\) the index afterward. The ratio cancels out the schooling already completed:

\frac{h_1}{h_0} = \frac{e^{r(s+3)}}{e^{rs}} = e^{3r} = e^{3(0.06)} = e^{0.18} = 1.1972

The new index is \(1.1972\) times the old index. The implied productivity-level increase is \((1.1972-1)\times 100=19.72\%\), or 19.72% to two decimal places under this model.

Check: An index of 1.0000 becomes 1.1972; the difference is 0.1972, which is 19.72% of the original index. This is a level comparison after schooling is completed, not a claim that productivity grows by 19.72% every year. The example avoids treating a cumulative schooling return as an annual growth rate.

From Individual Skills to Aggregate Productivity

An education program may reach only part of the workforce. Its aggregate effect therefore depends both on the productivity difference between trained and untrained workers and on the change in the share trained. In a simple calculation, average effective labor per worker is a weighted average of the productivity indices of the groups. To translate this into output, the production function matters: with diminishing returns to effective labor, the percentage change in output can be smaller than the percentage change in effective labor.

Worked Example: Scaling Up a Vocational Training Program

Method: workforce weighting followed by a production-function calculation. A hypothetical economy has two worker groups. Untrained workers have a productivity index of 1.00, and trained workers have an index of 1.30. The trained share rises from 20% to 50%. Assume total employment and physical capital are fixed during the comparison, technology is unchanged, and output follows \(Y=A K^{0.4}(HL)^{0.6}\), where \(H\) is average human capital. Find the percentage increase in output at the new workforce composition.

At the initial composition, average human capital is \(0.80(1.00)+0.20(1.30)=0.80+0.26=1.06\). At the new composition, it is \(0.50(1.00)+0.50(1.30)=0.50+0.65=1.15\). Thus the human-capital ratio is \(1.15/1.06=1.084906\), a rise of about 8.49%.

With capital, technology, and employment fixed, output varies with \(H^{0.6}\). The output ratio is therefore:

\frac{Y_1}{Y_0} = \left(\frac{1.15}{1.06}\right)^{0.6} = 1.0501

Output rises by \((1.0501-1)\times100=5.01\%\), to two decimal places. The result is a productivity-level comparison under the stated short-run assumptions. It does not imply that output will continue growing 5.01% faster each year.

Check: Because the output elasticity of human capital is 0.6, the 8.49% increase in average human capital maps to a smaller output increase. Applying the 8.49% directly to output would ignore the production function. This example avoids that common aggregation and elasticity error.

When Education Can Affect Sustained Growth

A one-time rise in educational attainment can raise human capital and output, but a lasting increase in growth requires more. For example, a model may allow human capital to keep accumulating through ongoing education, training, or learning in work. Alternatively, higher education may improve the effectiveness of research effort, producing a continuing increase in the rate of technological innovation. These are distinct mechanisms; higher schooling by itself does not guarantee either one.

A simple reduced-form innovation relationship makes the second channel clear. Let \(g_A=\eta R\), where \(g_A\) is the annual growth rate of technology, \(R\) is the amount of effective research input, and \(\eta\) is research productivity. Suppose education permanently raises researchers’ capabilities and hence \(\eta\), while the amount of research input remains unchanged. In this model, the growth rate of technology rises for as long as the higher research productivity applies.

This reduced-form expression is an assumption, not a universal law. The outcome depends on research skills being useful, the innovation process responding to them, and resources being allocated to research. If education raises researchers’ output only during a temporary transition, it can generate a temporary growth advantage rather than a permanently higher growth rate.

Worked Example: Higher Education and Research Productivity

Method: compare growth rates, then compound the growth-rate difference. In a hypothetical economy, technology grows at 2.0% per year under the initial education system. A higher-education reform permanently raises research productivity by 20%, while the quantity of research input is fixed. Assume the reduced-form relationship \(g_A=\eta R\) applies, the baseline growth rate is entirely explained by this relationship, and output is proportional to technology over the comparison period. Calculate the new annual technology growth rate and the technology advantage after 15 years relative to the unreformed path.

Because \(R\) is fixed, a 20% increase in \(\eta\) raises \(g_A\) by 20%: \(2.0\%\times1.20=2.4\%\) per year. The difference from the initial growth rate is \(2.4\%-2.0\%=0.4\) percentage points per year, or \(0.004\) in decimal form.

The ratio of technology under reform to technology without reform after 15 years is found by compounding the difference in continuously compounded growth rates:

\frac{A_{15}^{\text{reform}}}{A_{15}^{\text{baseline}}} = e^{(0.024-0.020)15} = e^{0.06} = 1.0618

The technology level is therefore about \((1.0618-1)\times100=6.18\%\) higher after 15 years, to two decimal places, under the assumptions. The annual growth rate is 0.4 percentage points higher throughout the period; 6.18% is the resulting accumulated level advantage, not the annual growth-rate increase.

Check: If both paths instead grew at 2.0%, their ratio would remain one. The higher path results from the extra 0.4 percentage points each year. The example avoids confusing a percentage increase in research productivity with a percentage-point increase in the growth rate.

Why the Level-versus-Growth Distinction Holds

The distinction follows directly from the growth-accounting equation. Suppose policy raises \(h\) from \(h_0\) to \(h_1\), with \(h_1>h_0\), while \(A\), \(k\), and the production shares do not change at the moment of comparison. Output per worker changes by the ratio below. Once the new human-capital level is reached, if \(h\) then grows at the same rate as before, its contribution to the growth rate is no larger than before.

\frac{y_1}{y_0} = \left(\frac{h_1}{h_0}\right)^{1-\alpha} \qquad\text{and}\qquad g_y = g_A+\alpha g_k+(1-\alpha)g_h

The first expression is a level comparison; the second describes ongoing growth under the model. During the years when a policy is raising educational attainment, human capital may grow temporarily faster, adding to growth during that transition. When the higher attainment has been achieved and \(g_h\) returns to its prior rate, that temporary growth contribution ends. The economy can remain on a higher output path without having a permanently higher growth rate.

By contrast, if education permanently raises \(g_h\), or raises \(g_A\) through a continuing innovation mechanism, the growth-accounting equation permits a sustained increase in \(g_y\), all else equal. Whether such a mechanism is plausible is an empirical and modelling question. A policy announcement or a higher average number of years in education is not, on its own, evidence that the long-run growth rate has changed.

Policy Design and Common Traps

Different forms of education may work through different channels. Basic schooling can build foundational literacy, numeracy, and learning capacity. Vocational training can develop task-specific skills or help workers use particular equipment. Higher education can support specialized professions and research. The economic effects depend on fit: training that does not match useful tasks may have a small productivity effect, while skills that complement technology or other workers may have larger effects.

Evaluation also requires attention to timing and costs. Students may spend time in education rather than working, so measured output can be lower during the transition even if future productivity improves. Public resources used for education have alternative uses. In addition, workers who pursue more schooling may differ from those who do not, so a simple comparison of their earnings or productivity does not necessarily identify the causal effect of schooling. These considerations do not negate human-capital benefits; they limit what can be inferred from a single before-and-after or group comparison.

Do not call a higher level a higher growth rate. A one-time rise in \(h\) can permanently raise the output path without permanently changing annual growth.
Do not ignore transition timing. Education may temporarily withdraw students from current production while building skills for later use.
Do not treat every worker as equally affected. Aggregate effects depend on coverage, skill gains, and how skills are used.
Do not infer causation from an association alone. Education choices and productivity may both depend on other factors.
Do not assume innovation responds automatically. A sustained growth effect requires an ongoing human-capital or innovation mechanism in the model.

Key takeaway: Education policy can raise productivity by increasing workers’ human capital. A one-time increase in skill attainment generally creates a higher output level, while a permanently higher growth rate requires a lasting change in human-capital accumulation or another continuing productivity mechanism, such as research effectiveness.

Check Your Understanding

For each hypothetical case, distinguish the relevant productivity or growth effect and calculate the requested quantity.

A training program raises the trained share of workers from 10% to 40%. Trained workers have a productivity index of 1.25 and others have an index of 1.00. Under a linear effective-labor average, calculate the percentage change in average human capital. (Answer: 7.32% increase, from 1.025 to 1.100.)
Assume the human-capital index is \(h=e^{0.05s}\), where \(s\) is years of schooling. A policy adds two completed years, with all other conditions fixed. What is the human-capital level increase, and is it automatically an annual growth-rate increase? (Answer: \(e^{0.10}-1=10.52\%\); it is a level increase, not automatically an annual growth-rate increase.)
Output per worker follows \(y=A k^{0.3}h^{0.7}\). Education raises \(h\) by 4%, while \(A\) and \(k\) are fixed. Calculate the percentage change in output per worker, rounding to two decimals. (Answer: \(1.04^{0.7}-1=2.78\%\) increase.)
A reduced-form innovation model has \(g_A=\eta R\). The initial technology growth rate is 1.5% per year. Education raises \(\eta\) by 12% permanently, with \(R\) fixed. Find the new growth rate and the technology-level advantage after 10 years relative to the initial path, using continuous compounding. (Answer: 1.68% annual growth; \(e^{(0.0168-0.015)10}-1=1.82\%\) technology-level advantage.)
Suppose \(g_y=g_A+0.3g_k+0.7g_h\). A continuing education system raises \(g_h\) from 0.5% to 1.0% per year, while \(g_A\) and \(g_k\) are unchanged. By how many percentage points does output-per-worker growth rise? (Answer: \(0.7(1.0\%-0.5\%)=0.35\) percentage points per year.)

What You'll Learn