Romer's model breaks the exogeneity of technological progress by making the rate of innovation endogenous to economic incentives, particularly R&D investment. The model features a separate R&D sector where firms create new varieties of intermediate goods. Because ideas have public good properties and imperfect excludability, the economy can sustain positive long-run growth without population growth, with sustained increases in living standards driven by intentional innovation rather than exogenous technological manna.
From your study of steady-state growth analysis and market failures, you know two things that set up Romer's contribution. First, in the Solow model, long-run growth in output per worker comes entirely from technological progress — but that progress is assumed to fall from the sky at a constant rate, with no explanation of where it comes from or why it varies across countries. Second, you know that externalities cause markets to deviate from social optimality. Romer's 1990 model connects these ideas: technological progress is the result of deliberate, profit-motivated investment in research, and because knowledge has externality-like properties, the market produces a suboptimal amount of it.
The model divides the economy into three sectors. The final goods sector uses labor and a variety of intermediate inputs to produce output, with a production function that exhibits diminishing returns to each individual input but constant returns overall. The intermediate goods sector consists of monopolistically competitive firms, each producing a unique variety of intermediate good using a patented design. The R&D sector employs researchers who combine existing knowledge with their own effort to produce new designs — blueprints for new intermediate good varieties. When a new design is invented, it is patented, and the inventor earns monopoly profits from licensing it to an intermediate goods producer. These expected profits are what motivate R&D investment in the first place.
The crucial economic property of ideas is non-rivalry: using a blueprint to produce one unit of an intermediate good does not prevent someone else from using the same blueprint simultaneously. This distinguishes ideas from physical capital — a machine can only be in one factory at a time, but a design can be replicated infinitely at near-zero marginal cost. Non-rivalry means that the production function for the economy as a whole exhibits increasing returns to scale when you include knowledge alongside labor and capital. This is what breaks the Solow model's prediction of convergence: countries that invest more in R&D generate more ideas, which raise productivity, which funds more R&D, sustaining growth indefinitely without the diminishing returns that eventually choke off capital accumulation.
However, non-rivalry creates a problem: if ideas were also non-excludable (freely available to everyone), no firm could earn a return on R&D investment, and no one would bother innovating. Romer resolves this with partial excludability through patents — innovators get temporary monopoly rights over their designs, earning enough profit to justify the R&D cost, even though the knowledge eventually diffuses. The policy implications are profound. Because private R&D decisions do not account for the positive spillovers that new knowledge creates for future researchers, the market equilibrium involves too little innovation relative to the social optimum. This provides a rigorous justification for R&D subsidies, patent protection, and public funding of basic research — not as ad hoc interventions but as corrections for a well-defined market failure at the heart of economic growth.