2.1. The Neoclassical Model: Capital Accumulation and Exogenous Growth

The Neoclassical growth model serves as the foundational starting point for all modern growth analysis. Developed independently by Solow (1956) and Swan (1956), its primary focus is on capital accumulation. The model’s central insight comes from the principle of diminishing marginal productivity: as an economy accumulates more capital relative to labor, each additional unit of capital generates a progressively smaller increase in output. This simple but powerful assumption leads to profound, though ultimately limited, conclusions about the determinants of long-run economic growth, making it an indispensable benchmark against which all other theories are measured. The Neoclassical model’s failure to explain the origin of technological progress led economists to develop theories of endogenous growth, which we will explore in subsequent sections.

2.1.1. The Solow-Swan Model

In its simplest form, without population growth or technological progress, the Solow-Swan model describes an economy where output depends solely on the capital stock. The production function is given by Y = F(K), which is assumed to have positive but diminishing marginal returns to capital (satisfying conditions 1.1 and 1.2). Net investment, or the change in the capital stock (ΔK), is determined by the portion of output that is saved (sY) minus the portion of the existing capital stock that depreciates (δK). This fundamental dynamic is expressed as:

ΔK = sF(K) – δK

This equation implies that the economy will converge toward a point where investment is just enough to cover depreciation.

Steady State: The economy reaches a “steady state” (K*) when the capital stock is no longer changing, i.e., when net investment is zero (ΔK = 0). At this point, total savings exactly offset total depreciation: sF(K*) = δK*. In the absence of technological progress, this means that total output also becomes constant, and economic growth ceases.

Introducing Population Growth

When we introduce population growth at a constant rate n, the logic remains the same, but the analysis shifts to per capita quantities. The crucial variable becomes the capital-labor ratio, k = K/L. For this ratio to remain constant, investment must now cover not only the depreciation of capital (δk) but also provide new capital for the growing workforce (nk). The dynamic equation for the capital-labor ratio becomes:

Δk = sf(k) – (n + δ)k (Equation 1.7)

As illustrated in Figure 1.2 from the source text, the savings per worker curve, sf(k), is concave due to diminishing returns, while the “capital-widening” line, (n + δ)k, is linear. The economy converges to a steady-state capital-labor ratio, k*, where the two curves intersect. At this point, output per person (y = f(k*)) is constant, and per capita growth is zero.

Introducing Exogenous Technological Change

The model can only explain the persistent, long-run growth observed in modern economies by introducing exogenous technological change, which continually offsets the effect of diminishing returns. Assuming technology is “labor-augmenting” and grows at a constant rate g, the production function can be written in Cobb-Douglas form as:

Y = (AL)^(1-α) * K^α (Equation 1.8)

Here, AL represents the number of “efficiency units” of labor, which grows at a rate n + g. The analysis now focuses on capital per efficiency unit, κ = K/(AL). The core dynamic equation becomes:

Δκ = sκ^α – (n + g + δ)κ (Equation 1.9)

The economy converges to a steady-state κ*, where capital per efficiency unit is constant. While output per efficiency unit is constant in this steady state, output per person (y = Y/L = Aκ^α) grows. The model’s ultimate conclusion is that the long-run growth rate of output per person is determined solely by the exogenous rate of technological progress, g. Policy changes, like an increase in the savings rate, can raise the level of the growth path but cannot permanently alter its slope (the long-run growth rate).

This framework gives rise to the concept of conditional convergence.

Conditional Convergence: The prediction that two countries will converge to the same steady-state level of income per capita if and only if they share the same technology (as defined by A_t) and the same fundamental parameters (s, n, δ) that determine capital accumulation. If these conditions differ, they will converge to different steady states, and therefore not to each other.

2.1.2. The Cass-Koopmans-Ramsey Model: Optimal Savings

The Cass-Koopmans-Ramsey (CKR) model is a crucial extension of the Neoclassical framework that endogenizes the savings rate. Instead of assuming a constant savings rate s, the CKR model derives savings from the intertemporal optimization choices of a representative household seeking to maximize its lifetime utility.

The household’s problem is to choose a path of consumption and capital to maximize its lifetime utility subject to the economy’s capital accumulation constraint:

K_t+1 = F(K_t) + (1-δ)K_t – c_t (Equation 1.11)

The solution to this problem is characterized by the Euler equation (Equation 1.15), which is the condition for optimal consumption smoothing over time. The intuition behind this equation is that, at the optimum, the household is indifferent between consuming one more unit today versus saving it, investing it, and consuming the proceeds tomorrow. It is the point where the marginal rate of substitution between consumption today and tomorrow must equal the marginal rate of transformation (the return on capital). The canonical form of the Euler equation is:

r = ρ + εg (Equation 1.19)

where r is the net rate of return on capital, ρ is the household’s rate of time preference (impatience), g is the growth rate of consumption, and ε is the inverse of the intertemporal elasticity of substitution (a measure of how resistant the household is to consumption fluctuations).

When exogenous technological change at rate g is incorporated into the CKR model, the central result of the Solow-Swan model is confirmed. In the steady state, capital, consumption, and output all grow at the exogenous rate of technological progress, g, provided the transversality condition ρ + (1 – ε)g > 0 holds.

The Neoclassical model’s starkest prediction, and its primary weakness, is its inability to explain the origin of the technological progress that is the sole driver of long-run growth. This limitation motivates the shift toward endogenous growth theories.

2.2. The AK Model: The Dawn of Endogenous Growth

The AK model represents the first major theoretical attempt to endogenize long-run growth, meaning to explain it from within the economic system rather than relying on an external, unexplained force. The Neoclassical model’s failure to explain the origin of ‘g’ led economists to develop theories of endogenous growth, the simplest of which is the AK model. It achieves this by making a critical departure from the Neoclassical framework: it eliminates the assumption of diminishing returns to capital. This is often justified by arguing that as firms accumulate physical capital, they also generate positive externalities through “learning by doing,” which creates new knowledge and offsets the tendency for the marginal product of capital to fall.

The basic structure of the AK model is strikingly simple. The aggregate production function is linear in the capital stock:

Y = AK

Here, A is a constant representing the economy’s overall productivity level, and K is a broad measure of capital that includes physical, human, and knowledge capital. With capital accumulating according to the standard equation ΔK = sY – δK, the economy’s growth rate (g) can be derived as:

g = sA – δ

This equation has a direct and powerful implication: the long-run growth rate is endogenous. Policies or behaviors that increase the national saving rate s or improve aggregate productivity A will lead to a permanent increase in the long-run growth rate.

This brings us to the central debate in early endogenous growth theory. While the AK model could explain the reality of sustained, positive growth, Neoclassical advocates pointed out that it failed to explain the widely observed phenomenon of conditional convergence. As shown in Figure 2.1 from the source, which plots convergence patterns among U.S. states, poorer regions do tend to grow faster than richer ones, a fact consistent with the diminishing returns of the Neoclassical model but at odds with the basic AK model, which predicts a constant growth rate regardless of the initial capital stock.

A clever extension by Acemoglu and Ventura (2002) shows that even AK-type models can exhibit convergence in an open-economy setting. Their core argument is that international trade creates a terms-of-trade effect. If one country grows much faster than its trading partners, the global supply of its goods increases, causing their relative price to fall. This decline in the terms of trade lowers the effective return to capital (A) in that country, thereby slowing its growth rate until it converges with the rest of the world. In this version, convergence occurs not because of a diminishing physical marginal product of capital, but because of a diminishing value of the marginal product of capital.

Ultimately, the AK model’s primary weakness is its failure to explicitly distinguish between the mechanisms of capital accumulation and technological progress. This limitation set the stage for the next wave of endogenous growth theories: the innovation-based models.

2.3. Innovation-Based Growth Part I: The Product-Variety Model

The AK model’s failure to distinguish between capital accumulation and technological progress led economists to develop more sophisticated frameworks. This section covers the first of two major “innovation-based” endogenous growth theories. Developed by Romer (1990), the product-variety model formalizes the idea that long-run growth is sustained by increased specialization. In this framework, growth is driven by an expanding variety of specialized intermediate products used in production, rather than by the sheer accumulation of capital.

The model’s core production function for the final good is:

Y = L^(1-α) * ∫(from 0 to M_t) (x_i^α) di (Equation 3.2)

Here, L is labor, x_i is the quantity of intermediate input i, and M_t is the crucial variable representing the total variety (or number) of available intermediate products. In this setup, the measure of product variety M_t functions as the economy’s aggregate productivity parameter. As M_t increases, a given amount of resources can be spread more efficiently across a wider range of specialized inputs, increasing final output.

The key to endogenizing growth is explaining what drives the increase in M_t. New product varieties are created by profit-seeking entrepreneurs who invest in R&D. The incentive for this innovation is the prospect of earning perpetual monopoly rents from the sale of their unique intermediate product. For a version of the model where R&D uses labor, the equilibrium growth rate is:

g = (αλL – ρ) / (1 + αε)

where λ is R&D productivity, L is labor supply, ρ is the rate of time preference, α is the capital share, and ε is the inverse intertemporal elasticity of substitution.

This equation reveals the model’s prediction of a “scale effect”: the economy’s growth rate g is positively related to the size of the population L. This implies that larger economies should grow faster. However, this prediction has been challenged empirically. Jones (1995b) pointed out that the number of researchers in the U.S. has increased dramatically over the past several decades without a corresponding trend increase in the national growth rate, casting doubt on this simple scale effect.

The product-variety model’s primary limitation is its assumption that old products are never rendered obsolete. In this framework, the exit of a firm is detrimental to growth because it reduces product variety (M_t), which is the sole determinant of aggregate productivity. This stands in sharp contrast to a great deal of empirical evidence suggesting that firm turnover, exit, and the process of “creative destruction” are, in fact, crucial components of a dynamic, growing economy.

The limitations of the product-variety model, particularly its inability to account for obsolescence, motivate the need for a theory that explicitly incorporates creative destruction, leading directly to the Schumpeterian paradigm.

2.4. Innovation-Based Growth Part II: The Schumpeterian Model

The Schumpeterian paradigm, named for the economist Joseph Schumpeter, represents the final core growth theory we will examine. Its central mechanism is creative destruction.

Creative Destruction: The process where new, quality-improving innovations render old products, technologies, and firms obsolete. This focus on vertical (quality) innovation, as opposed to the horizontal (variety) innovation of the Romer model, provides a distinct and powerful analytical lens for understanding the dynamics of modern growth, particularly the roles of competition, firm turnover, and policy.

In a basic one-sector version of the model, the production function is given by:

Y_t = A_t^(1-α) * L^(1-α) * x_t^α

Here, A_t is the productivity parameter attached to the latest, leading-edge technology. Entrepreneurs spend resources R_t on R&D in the hopes of discovering a new technology. A successful innovation, which occurs with probability μ_t, multiplies the existing productivity parameter by a factor γ > 1.

The amount of R&D undertaken is determined by a “research arbitrage” equation, φ'(n_t) = π_L. The intuition is that, in equilibrium, firms will invest in R&D up to the point where the marginal cost of research must equal the marginal benefit (the increased probability of innovating multiplied by the value of the innovation).

The model’s central finding regarding long-run growth is summarized in Proposition 1:

Proposition 1: The economy’s average growth rate equals the frequency of innovations times the size of those innovations: g = μ(γ-1).

This simple and elegant result leads to several important comparative static implications for what drives growth:

• Productivity of Research (λ): A higher productivity of R&D increases the frequency of innovations and thus raises the average growth rate.
• Size of Innovations (γ): A larger step-size for each innovation boosts growth. This provides a formal basis for Gerschenkron’s “advantage of backwardness”: a country further behind the technological frontier can achieve larger productivity gains (γ is bigger) by adopting frontier technologies, allowing for faster catch-up growth.
• Property-Rights Protection: Stronger protection of intellectual property increases the monopoly profits (π_L) that reward innovation, encouraging more R&D and faster growth.
• Product Market Competition: In this basic model, increased competition lowers monopoly profits, which is predicted to reduce R&D incentives and slow growth. Note the subtle but important difference from later models: this initial prediction is at odds with recent empirical evidence, a point we will revisit in Module 4.
• Population Size (L): Like the product-variety model, this framework also predicts a scale effect, where a larger population leads to faster growth.

The questionable “scale effect” can be eliminated from the Schumpeterian model by incorporating an insight from Young (1998). If, as population grows, there is a proliferation of different product lines, research efforts get spread more thinly across this expanding set of products. This dilution of R&D effort per product line can offset the aggregate scale effect, making the overall growth rate independent of population size in the long run.

The Schumpeterian model provides a flexible and powerful framework for analyzing the nuanced processes of technological change, competition, and policy. It will be extended and applied throughout the remainder of these notes to understand the deeper determinants of growth.