Economic growth models: Harrod-Domar and Solow

In short

Growth models explain how saving, investment, population and technology determine the long-run rate at which an economy's output grows. The Harrod-Domar model (1939-46) treats growth as driven by saving and a fixed capital-output ratio, and its instability ("knife-edge") result made it the workhorse of Indian planning from the Second Five Year Plan onwards; the Solow-Swan model (1956) replaces fixed coefficients with a smooth, substitutable production function and shows that saving alone cannot raise the long-run growth rate — only technical progress can. The topic is examined in BA/MA Macroeconomics papers (CBCS core course on Macroeconomics/Growth), in UGC-NET and GATE Economics (XH-C1), and heavily in the UPSC Economy paper and the RBI Grade B and NABARD/SEBI Grade A exams. Employers test it through ICOR arithmetic, growth accounting and TFP questions — in NITI Aayog and RBI research work, in rating agency and bank economics desks, and in consulting macro briefs.

Explained

Setting up: the growth arithmetic common to both models

• Output (Y) is produced with capital K and labour L. Saving S = sY, and in a closed economy with no government S = I (investment).
• Gross investment covers the net addition to the capital stock plus replacement of worn-out capital: I = ΔK + δK, where δ is the depreciation rate. Net investment is the change in the capital stock itself, ΔK = gross investment minus depreciation.
• Capital-output ratio: the average ratio v = K/Y; the incremental capital-output ratio (ICOR) = ΔK/ΔY = investment needed per unit of extra output per year. India's ICOR has typically run around 4-5 in recent decades (RBI and Economic Survey estimates vary by method and year).
• Per-capita terms: k = K/L, y = Y/L. If output grows at g_Y and population at n, per-capita output grows at approximately g_Y − n.
• Growth rate of a product: if Y = AK^αLβ, then g_Y = g_A + α·g_K + β·g_L. This decomposition is the basis of growth accounting.

The Harrod-Domar model

• Assumptions: fixed-coefficient (Leontief) technology, so capital and labour are not substitutable; constant saving ratio s; constant capital-output ratio v; no depreciation in the simplest version; closed economy; labour surplus, so capital is the only binding constraint.
• Domar's arithmetic: investment has a dual character. On the demand side, through the multiplier, ΔY_d = ΔI/s. On the supply side, it adds capacity: ΔY_s = σI, where σ = 1/v is the output-capital ratio (Domar's "potential social average productivity of investment"). Equating the two gives ΔI/I = sσ = s/v: investment must grow at s/v for capacity to stay fully used.
• Harrod's three growth rates:
  1. Actual growth rate G = s/C, where C is the realised incremental capital-output ratio.
  2. Warranted growth rate G_w = s/C_r, where C_r is the required capital-output ratio consistent with entrepreneurs' intentions — the rate at which the economy must grow to keep capital fully employed and expectations satisfied.
  3. Natural growth rate G_n = n + m (labour force growth plus labour productivity growth) — the maximum sustainable rate, set by population and technology; also called the full-employment growth rate.
• Knife-edge instability: if G > G_w, realised C falls below C_r, capital looks scarce, firms invest more, and growth accelerates away upward (inflationary boom). If G < G_w, capital appears excess, investment is cut, and the economy spirals into deeper recession. Equilibrium growth is therefore unstable — a razor's edge.
• Long-run outcomes: G_w > G_n means the warranted rate cannot be achieved for want of labour — chronic depression and secular stagnation; G_w < G_n means labour is abundant relative to capital — chronic unemployment and, with easy finance, chronic inflation. Steady full-employment growth needs G = G_w = G_n, which nothing in the model guarantees.
• Policy reading: to grow faster, raise s (saving and investment rate) or lower v (use capital more productively). This gave planners the simple target rule *g = s/v*.
• Harrod-Domar in Indian planning: the Second Five Year Plan (1956-61) and the Mahalanobis two-sector/four-sector model used exactly this logic — raise the investment rate, direct it to Sector K (capital goods) so that the economy's future capacity to invest rises, accepting slower immediate consumption growth. The investment-rate target and ICOR remained central to all Plan documents up to the Twelfth Plan (2012-17); the Planning Commission was replaced by NITI Aayog in 2015 and formal five-year plans ended, but ICOR-based arithmetic still appears in Economic Survey and RBI analysis.
• Limitations: no factor substitution (the fixed v is the key unrealistic assumption); s and v treated as exogenous constants; ignores prices, interest rates and relative factor scarcity; no explicit technical progress; no role for institutions, human capital or the external sector; treats capital as the sole constraint, which mis-describes economies constrained by skills, infrastructure, demand or foreign exchange; the instability result is not borne out by actual OECD or Indian experience.

The Solow-Swan neoclassical model

• Assumptions: one good; production function Y = F(K, AL) with constant returns to scale, positive but diminishing marginal products, and smooth substitution between K and L; competitive markets; constant saving rate s; labour force grows at n; depreciation δ; technology A grows exogenously at g (labour-augmenting/Harrod-neutral); closed economy. Usually Cobb-Douglas: Y = A K^α L^1−α with α ≈ 1/3 (capital's income share).
• Intensive form: dividing by L, y = f(k) with f'(k) > 0, f''(k) < 0 — the hallmark of diminishing returns to capital.
• Fundamental dynamic equation (no technical progress): Δk = s·f(k) − (n + δ)k. Actual investment per worker minus "required" or break-even investment (what is needed to replace worn-out capital and to equip new workers).
• Steady state: where Δk = 0, i.e. s·f(k\*) = (n + δ)k\*. At k\*, capital per worker, output per worker and consumption per worker are constant; total Y, K and C grow at n. Per-capita growth is zero.
• Stability: unlike Harrod-Domar, the steady state is globally stable. If k < k\*, sf(k) > (n+δ)k so k rises; if k > k\*, k falls. Diminishing returns do the stabilising work — the knife-edge disappears because v = K/Y is endogenous and adjusts.
• With technical progress at rate g: variables are measured per effective worker (k̃ = K/AL). Then Δk̃ = s·f(k̃) − (n + g + δ)k̃, and in steady state:
  • output per worker and capital per worker grow at g;
  • total output and capital grow at n + g;
  • the capital-output ratio, the rate of return and factor shares are constant — consistent with Kaldor's stylised facts.
• Central result: the long-run per-capita growth rate equals g, the rate of technical progress, and is independent of s. A higher saving rate raises the *level* of the steady-state income path (a one-off, permanently higher level) and gives faster growth only during the transition — a level effect, not a growth effect. Since g is exogenous and unexplained, Solow growth is often called "growth without a theory of growth".
• Comparative statics: ↑s → higher k\*, y\* ; ↑n → lower k\*, y\* (capital widening absorbs saving) ; ↑δ → lower k\*, y\*. Technical progress is the only source of permanent per-capita growth.
• Capital deepening vs capital widening: deepening raises K/L (raises y, subject to diminishing returns); widening only equips extra workers at the existing k.
• Golden rule: the saving rate that maximises steady-state consumption per worker satisfies f'(k_g) = n + g + δ. For Cobb-Douglas this means s_golden = α. If k\* > k_g the economy over-saves (dynamically inefficient — consumption can rise now and forever by saving less); if k\* < k_g, raising s costs consumption today to gain more later.
• Convergence: because poor countries start with low k and high marginal product of capital, they should grow faster. Absolute convergence (all countries to the same level) is rejected by data; conditional convergence — convergence to each country's *own* steady state, given its s, n, δ and institutions — is broadly supported, with an estimated speed of roughly 2% of the income gap closed per year (Barro, Mankiw-Romer-Weil), which is slower than the basic model predicts and is one reason human capital was added (α for broad capital ≈ 0.6-0.8).
• Growth accounting and the Solow residual: g_Y = g_A + α·g_K + (1−α)·g_L. The part not explained by measured inputs, g_A = TFP growth, is the "Solow residual" — the "measure of our ignorance". Solow's 1957 US study attributed the bulk of per-capita growth to TFP, not capital accumulation. For India, studies (RBI, NITI Aayog, Penn World Table based work) typically find TFP contributing roughly a quarter to a third of growth in the post-1991 period, with capital accumulation dominant; estimates differ substantially by method and vintage, so quote them as ranges.
• Limitations: technical progress is exogenous and unexplained; no human capital, R&D, institutions or policy in the basic version; assumes a closed economy with no capital flows; a single aggregate good and perfect competition; cannot explain large persistent cross-country income gaps or sustained divergence; constant s is not optimising behaviour (fixed by Ramsey-Cass-Koopmans, which derives saving from utility maximisation).
• What came after: endogenous growth theory — AK models (Rebelo), human capital (Lucas, Mankiw-Romer-Weil augmented Solow), R&D and ideas with increasing returns (Romer 1990), Schumpeterian creative destruction (Aghion-Howitt) — makes g itself a function of saving, education and innovation policy, so policy can change the long-run growth rate.

Harrod-Domar versus Solow compared

• Technology: fixed coefficients vs smooth substitution — the single difference from which everything else follows.
• Capital-output ratio: exogenous constant vs endogenous, adjusting to equate saving and required investment.
• Stability: knife-edge instability vs a globally stable steady state.
• Role of saving: permanently raises growth (g = s/v) vs raises only the income level, not long-run growth.
• Source of long-run per-capita growth: more investment vs exogenous technical progress.
• Tradition: Keynesian (demand-side, short-run multiplier logic extended to the long run) vs neoclassical (supply-side, full employment assumed).
• Use: Harrod-Domar for quick planning arithmetic and investment-requirement estimates; Solow for growth accounting, convergence analysis and cross-country comparison.

The Indian context

• Investment and saving rates: India's gross domestic saving and gross fixed capital formation rates (as a share of GDP) peaked in the mid-2000s boom at well above 30% and have since been lower; current figures are published annually by MoSPI in the National Accounts Statistics and summarised in the Economic Survey — quote the latest release rather than a remembered number.
• ICOR as a policy diagnostic: a rising ICOR signals that each rupee of investment buys less output — attributed to stalled projects, infrastructure gaps and poor capital allocation. Falling ICOR with a given saving rate is the cheapest route to faster growth under *g = s/v*.
• Harrod-Domar's planning legacy: Second Plan's heavy-industry strategy, licensing of capital goods capacity, public-sector investment in steel and power; its weakness — assuming capital was the only constraint — is part of the standard critique of the pre-1991 regime, where foreign exchange and skills were binding too.
• Solow's reading of India: post-1991 acceleration is explained partly as transition dynamics (capital deepening from a low k after reform raised the effective s and the return to capital) and partly as TFP gains from trade opening and deregulation; the sustainability of 7-8% growth is debated precisely in terms of whether it rests on accumulation or on TFP.
• Related policy frames: "demographic dividend" is the n-term in Solow working favourably for a few decades; Production Linked Incentive schemes, National Education Policy 2020 and the Atal Innovation/R&D push are endogenous-growth-style interventions aimed at g, not at s.

Key points

• Harrod-Domar growth rate - g = s/v: growth equals the saving rate divided by the capital-output ratio.
• ICOR - ΔK/ΔY, the investment needed for one extra unit of annual output; India's typically around 4-5.
• Warranted rate G_w - s/C_r, the growth rate that keeps entrepreneurs' capital fully and willingly employed.
• Natural rate G_n - n + m, the ceiling set by labour force growth and labour productivity growth.
• Knife-edge - any deviation of actual from warranted growth is self-amplifying, so equilibrium is unstable.
• G_w > G_n - secular stagnation; G_w < G_n - chronic unemployment and inflationary pressure.
• Fixed coefficients - Harrod-Domar's defining and most criticised assumption: no substitution between K and L.
• Solow fundamental equation - Δk = sf(k) − (n + δ)k; with technology, Δk̃ = sf(k̃) − (n + g + δ)k̃.
• Steady state - sf(k\*) = (n + g + δ)k\*; k̃ and ỹ constant, Y grows at n + g, Y/L grows at g.
• Diminishing returns to capital - the mechanism that makes the Solow steady state stable and ends growth from pure accumulation.
• Saving is a level effect - higher s permanently raises income, but long-run growth stays g.
• Golden-rule saving - f'(k_g) = n + g + δ; with Cobb-Douglas, s = α (capital's share, about one-third).
• Capital deepening vs widening - raising K/L vs merely equipping new workers at the same K/L.
• Conditional convergence - countries converge to their own steady states; measured speed roughly 2% of the gap a year.
• Growth accounting - g_Y = g_A + αg_K + (1−α)g_L; the unexplained g_A is TFP, the Solow residual.
• Kaldor facts - constant K/Y, constant return on capital, steady per-capita growth: reproduced by Solow with labour-augmenting technology.
• Endogenous growth - makes g a function of R&D, human capital and policy, curing Solow's main gap.
• Indian use - Second Plan and Mahalanobis strategy rest on Harrod-Domar; Economic Survey and RBI TFP studies use Solow.

Formulas and examples

Example 1 — Harrod-Domar target arithmetic. An economy saves 32% of GDP and has an ICOR of 4.0, with depreciation ignored.

• g = s/v = 0.32 / 4.0 = 0.08 = 8% per year.
• Population grows at 1.2%, so per-capita growth = 8 − 1.2 = 6.8%; income per head doubles in about 70/6.8 ≈ 10.3 years.
• Now suppose inefficiency raises ICOR to 5.0 with s unchanged: g = 0.32/5.0 = 6.4% — the same saving buys 1.6 percentage points less growth.
• To restore 8% growth at v = 5.0, the required investment rate is I/Y = g×v = 0.08 × 5.0 = 40% of GDP, a rise of 8 percentage points.
• With depreciation at 3% of the capital stock and v = 4: g = 0.32/4 − 0.03 = 0.08 − 0.03 = 5% per year.

Example 2 — Harrod's three rates and the knife-edge. s = 0.20, required capital-output ratio C_r = 4, labour force growth n = 2%, labour productivity growth m = 1.5%.

• G_w = 0.20/4 = 5%; G_n = 2 + 1.5 = 3.5%.
• G_w > G_n, so the warranted rate cannot be sustained: the economy runs out of labour before it runs out of capital — Harrod's case of secular stagnation, with actual growth held to 3.5% and realised growth chronically below warranted.
• If actual growth is 6% (above G_w = 5%), realised C = s/G = 0.20/0.06 = 3.33 < C_r = 4 → capital appears deficient → more investment → growth rises further. Instability confirmed.

Example 3 — Solow steady state, Cobb-Douglas. Y = K^1/3 L^2/3 (so A = 1, α = 1/3), s = 0.24, n = 0.01, g = 0, δ = 0.05.

• Steady state: s·k^1/3 = (n+δ)k → k\*^{1 − 1/3} = s/(n+δ) → k\*^2/3 = 0.24/0.06 = 4 → k\* = 4^3/2 = 8.
• y\* = k\*^1/3 = 8^1/3 = 2. Capital-output ratio = 8/2 = 4; check with s/(n+g+δ) = 0.24/0.06 = 4. ✓
• Consumption per worker = (1 − s)y\* = 0.76 × 2 = 1.52; break-even investment = 0.06 × 8 = 0.48 = s·y\* = 0.24 × 2 = 0.48. ✓
• Raise s to 0.36: k\*^2/3 = 0.36/0.06 = 6 → k\* = 6^1.5 ≈ 14.70; y\* = 14.70^1/3 ≈ 2.449 (= √6). Output per worker rises by about 22.5% — a permanent *level* gain, but long-run growth is still 0% per head because g = 0.
• Golden rule: s_golden = α = 1/3. At s = 1/3, k_g^2/3 = (1/3)/0.06 = 5.556 → k_g ≈ 13.09, y_g ≈ 2.357, c_g = (2/3)(2.357) ≈ 1.571 — the highest steady-state consumption. So s = 0.24 under-saves and s = 0.36 over-saves (c at s=0.36 is 0.64 × 2.449 ≈ 1.567 < 1.571).

Example 4 — Growth accounting / Solow residual. An economy records real output growth of 7.0%, capital stock growth of 9.0% and employment growth of 1.5%; capital's income share α = 0.35.

• Capital's contribution = 0.35 × 9.0 = 3.15 pp; labour's = 0.65 × 1.5 = 0.975 pp.
• TFP growth g_A = 7.0 − 3.15 − 0.975 = 2.875 pp, about 41% of total growth (2.875/7.0).
• Per-capita output growth = 7.0 − 1.5 = 5.5%; of this, capital deepening contributes α(g_K − g_L) = 0.35 × 7.5 = 2.625 pp and TFP 2.875 pp.
• Reading: a little under half the per-capita growth comes from accumulation (2.625 of 5.5 pp), which diminishing returns will erode; sustaining 7% requires TFP.

Example 5 — Investment requirement for a growth target (planning-style). Target real GDP growth 7.5%, ICOR 4.5, current GDP ₹300 lakh crore, population growth 0.9%.

• Required investment rate = 7.5 × 4.5 = 33.75% of GDP = 0.3375 × ₹300 lakh crore ≈ ₹101.25 lakh crore of investment in the year.
• If the domestic saving rate is 30%, the gap of 3.75% of GDP (≈ ₹11.25 lakh crore) must come from net foreign capital inflows — i.e. a current account deficit of that size.
• Per-capita growth = 7.5 − 0.9 = 6.6%; doubling time ≈ 70/6.6 ≈ 10.6 years.

Where it is used

• Government and policy research: NITI Aayog, Ministry of Finance (Economic Survey chapters on growth and investment), MoSPI National Accounts — investment-rate and ICOR arithmetic for growth projections; the Fifteenth Finance Commission and its successor use growth paths built this way.
• Central bank and financial regulators: RBI Department of Economic and Policy Research publishes potential-output, TFP and growth-accounting work; Monetary Policy Committee statements reference potential growth and the output gap, which rest on Solow-type supply-side estimates. Jobs: RBI Grade B (DEPR), NABARD, SEBI research.
• Rating agencies, banks and brokerages: CRISIL, ICRA, CARE, SBI Economic Research, bank treasury and equity strategy teams forecast medium-term GDP using saving-investment-ICOR frameworks and TFP assumptions for long-horizon valuation.
• Multilateral and consulting work: World Bank, IMF Article IV and ADB country reports on India use conditional convergence and growth decompositions; McKinsey/BCG India growth reports use the same accounting.
• Project and infrastructure appraisal: incremental capital-output logic underlies arguments for public capex (Gati Shakti, National Infrastructure Pipeline) — whether a rupee of capital expenditure buys more output than a rupee of transfers.
• Teaching, UPSC and academia: Economics Optional and GS Paper III, UGC-NET, GATE Economics, and university Development Economics courses where the Mahalanobis model and Second Plan strategy are assessed.

Interview questions

• Q: State the Harrod-Domar growth equation and what each term means. A: g = s/v, where s is the saving (and investment) rate as a share of national income and v is the capital-output ratio or ICOR; growth rises with saving and falls as each unit of output needs more capital.
• Q: What is the knife-edge problem? A: Harrod's warranted growth path is unstable: if actual growth exceeds the warranted rate, capital appears scarce and investment rises further, accelerating growth; if it falls short, excess capacity cuts investment and growth collapses — nothing pulls the economy back.
• Q: Distinguish the warranted and the natural growth rate. A: The warranted rate s/C_r is the rate that keeps entrepreneurs content with their capital stock; the natural rate n + m is the maximum the labour force and productivity allow. Steady full-employment growth needs them to coincide, which is not guaranteed.
• Q: Why does a higher saving rate not raise long-run growth in Solow? A: Diminishing returns to capital mean extra saving eventually goes entirely into replacing depreciation and equipping new workers; the economy settles at a higher level of income per head but the same long-run growth rate, g.
• Q: Then what does determine long-run per-capita growth in Solow? A: Only the exogenous rate of labour-augmenting technical progress, g — which the model assumes rather than explains, the gap endogenous growth theory fills.
• Q: Write the steady-state condition and the resulting capital-output ratio. A: sf(k̃) = (n + g + δ)k̃; dividing both sides by f(k̃) gives K/Y = s/(n + g + δ).
• Q: What is the golden-rule saving rate? A: The saving rate maximising steady-state consumption per worker, where the marginal product of capital equals n + g + δ; with a Cobb-Douglas function it equals capital's income share α, about one-third.
• Q: Difference between absolute and conditional convergence? A: Absolute convergence says all countries approach the same income level and is rejected by the data; conditional convergence says each country converges to its own steady state determined by its s, n, δ and institutions, and is broadly supported.
• Q: What is the Solow residual and why is it called a measure of our ignorance? A: It is the part of output growth left after weighting capital and labour growth by their income shares — attributed to TFP, but it also absorbs measurement error, quality change, reallocation and utilisation effects, so it is a residual rather than a measured quantity.
• Q: How did Harrod-Domar shape Indian planning? A: It gave the Second Five Year Plan its target rule — raise the investment rate and allocate it to capital goods (the Mahalanobis strategy) to raise future capacity — treating capital as the binding constraint on growth.
• Q: India's ICOR rises from 4 to 5 while the investment rate is unchanged at 32%. What happens? A: Growth falls from 8% to 6.4%; restoring 8% would need a 40% investment rate, so improving capital efficiency is the cheaper lever.
• Q: Is faster population growth good or bad for growth in these models? A: In Solow it lowers steady-state income per worker because saving must widen capital across more workers; it raises total output growth (n + g). In Harrod's framework a higher n raises the natural rate, which helps only if the warranted rate can keep up.
• Q: Which model would you use to forecast India's growth over ten years, and why? A: A Solow/growth-accounting frame, because it separates capital deepening (which diminishes) from TFP and labour-force effects; Harrod-Domar arithmetic is a useful cross-check on the investment needed for a given target.

Common mistakes

• Writing Harrod-Domar as g = s·v instead of g = s/v, or forgetting to subtract δ when the question specifies depreciation.
• Using the saving rate as a percentage with the ICOR in decimal form (or vice versa) — mixing 32 and 0.32 gives answers off by a factor of 100.
• Claiming a higher saving rate raises Solow's long-run growth rate; it raises only the level of the income path, with faster growth in transition only.
• Forgetting the technology term: in the Solow model with technical progress the break-even line is (n + g + δ)k̃, not (n + δ)k̃.
• Solving k\* from sk\*^α = (n+δ)k\* without raising to the power 1/(1−α) — the commonest algebra slip in steady-state questions.
• Treating the golden rule as the growth-maximising saving rate; it maximises steady-state consumption, and a higher s can leave the economy dynamically inefficient.
• Saying Solow predicts that all poor countries catch up — the model predicts convergence conditional on each country's own parameters.
• Confusing the average capital-output ratio K/Y with the incremental ratio ΔK/ΔY, and quoting one where the question needs the other.

Related topics

Consumption, investment and the multiplier · National income · Endogenous growth and human capital · Development economics and Indian planning · Inflation and unemployment · Fiscal policy and the Union Budget