Understanding Models of Information Diffusion: From Epidemic to Economic Theory

Every new idea, technology, or piece of information faces the same challenge: how does it move from a handful of early users to widespread adoption? Whether it is a farmer deciding to use a new variety of seed, a bank installing ATMs, or a factory switching to numerically controlled machines, the spread is rarely instant. It follows patterns. Researchers across sociology, economics, and statistics have built formal models to capture these patterns, and three families of models stand out: the epidemic model, economic models, and vintage and stock adjustment models. Each one explains the same phenomenon from a different angle, and together they give a far richer picture of how information and innovation travel through a society.

Why diffusion models matter 🔗

Diffusion is the process through which something of interest, an innovation, an idea, or information, spreads from one person or organisation to another. Understanding this process helps planners predict how quickly a new technology will be taken up, helps marketers time their campaigns, and helps policymakers decide where to intervene. The study of innovation diffusion is genuinely interdisciplinary, pulling together mathematics, physics, social science, marketing, and economics. The models below are the main tools used to make sense of it.

The epidemic model: when ideas spread like a contagion 🔗

The oldest and most intuitive approach treats the spread of an innovation exactly like the spread of a disease. Just as a virus passes from an infected person to a healthy one through contact, information about a new product passes from those who already use it to those who do not. This is why the approach is called the epidemic model, and it borrows its mathematics directly from biology.

The logic of contagion 🔗

In the early stage of diffusion, only a few users know about a new technology and can talk about its benefits. As more people adopt it, more sources exist to spread the word, so the rate of new adoption speeds up. The rate of adoption is proportional to both the number of existing adopters and the number of people who have not yet adopted. Past the midpoint, the pool of remaining non-adopters shrinks, contact between users and potential users becomes less likely, and the process slows down again.

This produces the famous S-shaped (sigmoid) curve of cumulative adoption: slow at first, rapid in the middle, and tapering off as the market saturates. Mathematically it is described by the logistic equation, the same equation used to model populations and outbreaks.

Strengths and limits of the epidemic view 🔗

The epidemic model is powerful because it matches what we actually observe in many markets, and the analogy extends naturally to modern settings such as the spread of rumours, fads, and “word of mouth” across social networks. Online, researchers have even adapted classic SIR (Susceptible-Infected-Recovered) disease models to track how topics spread through web forums and online discussions.

However, the model rests on a simplifying assumption: that everyone is equally likely to “catch” the innovation once exposed. In reality, potential adopters are not identical. Some have more money, some run larger operations, and some simply benefit more from adopting than others. The epidemic model treats people as passive carriers of information rather than as decision-makers weighing costs and benefits. This is exactly the gap that economic models set out to fill.

Economic models: market forces and firm size 🔗

Economic models start from a different premise. They assume that a firm adopts a new technology not simply because it has “heard” about it, but because adoption is profitable for that particular firm at that particular time. Diffusion happens gradually because firms differ from one another, and because the returns to adoption change as more firms join in. Economists have isolated several distinct forces, often summarised as rank, stock, and order effects, that are usefully contrasted with the epidemic effect in an influential body of empirical work.

The rank effect: profitability and firm characteristics 🔗

The rank effect says that firms can be ranked by how much they gain from a new technology. Those that benefit most adopt first; those that benefit least adopt last. What determines a firm’s rank? Its characteristics, including its size, the use of complementary technologies, and its managerial techniques. Studies of the intra-firm spread of computer numerically controlled machine tools in the United Kingdom found that profitability considerations mattered a great deal, with firm size and complementary technologies emerging as especially relevant, while the traditional epidemic effect received little support.

Why firm size matters 🔗

Firm size keeps appearing as a key driver, and for good reason. A larger firm can spread the fixed cost of a new technology over more output, can absorb more risk, and may have better access to finance. Research on the spread of ATMs in the UK financial sector found that firm size, growth, profitability, and learning were among the main factors shaping adoption. In some theoretical models, the relative size of firms even determines the order of adoption directly: in a market with one large firm and many small ones, the large firm adopts first and acts as a leader, with smaller firms following once uncertainty is resolved.

The order effect and strategic timing 🔗

The order effect introduces strategy. The benefit a firm gets from adopting may depend on how early it adopts relative to its rivals. An early adopter might capture scarce resources, a better market position, or a head start in learning, advantages that later adopters cannot fully recover. This turns adoption into a competitive game where timing itself carries value, which is why firms sometimes rush to adopt and sometimes deliberately wait.

Vintage and stock adjustment models 🔗

The third family looks more closely at the capital goods that embody a new technology and at how the existing stock of equipment shapes new purchases. These models are especially useful for explaining diffusion within industries, where machines are expensive, long-lived, and replaced only gradually.

The vintage model: technology in layers 🔗

The vintage model recognises that technology is embodied in physical capital, and that capital comes in different “vintages”, batches built in different years using the technology available at the time. A factory does not scrap all its old machines the moment a better one appears. Instead, new and old vintages coexist, and the firm modernises its capital stock only as it replaces worn-out or unprofitable equipment.

One vintage model of diffusion shows how learning-by-using affects this process: a vintage of intermediate age can be the most productive, because the very newest equipment has not yet matured while the oldest has exhausted its learning potential. Strong complementarity between vintages can slow down modernisation, because firms hesitate to replace equipment that still works well with what they already own. This explains why diffusion within an industry can take many years, with old and new technologies running side by side for a long time. The framework has been developed in detail in the academic literature on returns to diversity and learning-by-using.

The stock adjustment model and the stock effect 🔗

The stock adjustment model treats adoption as the gradual movement of an industry’s stock of a technology toward a desired or equilibrium level. The current stock of equipment adjusts toward this target over time, and the speed of adjustment depends on costs, capacity, and how far the actual stock sits below the desired one. This framing is common in studies of household appliances, where the share of households owning a device, the penetration ratio, follows a logistic path toward a saturation level, since a household rarely needs more than one washing machine, for example. In practice the vintage and stock approaches are often combined with epidemic-style logistic functions to project how stocks build up.

Closely related is the stock effect, which is one of the economic forces mentioned earlier. The idea is that the benefit of adopting falls as more firms have already adopted, because the extra output floods the market and pushes down prices or profits. The more “stock” of adopters there is, the weaker the incentive for the next firm to join. This naturally slows diffusion as adoption spreads, but for a reason that has nothing to do with information and everything to do with market returns.

How the models fit together 🔗

It is tempting to ask which model is correct, but that is the wrong question. Each captures a real mechanism. The epidemic model explains how information and awareness spread through contact. Economic models explain why some firms find adoption worthwhile before others, with firm size, profitability, and strategic timing all playing a part. Vintage and stock adjustment models explain why physical capital changes slowly and why the existing stock of equipment shapes new investment. Modern empirical work often tests rank, stock, order, and epidemic effects together, precisely because real diffusion is driven by all of them at once. For a student of information science, the lesson is that the movement of information is never just about communication; it is also about cost, profit, capital, and competition.

What do you think? Which of these models best explains how a recent technology spread through your own community or workplace, and could the same pattern of diffusion ever be deliberately engineered, or is it always shaped by forces outside any single planner’s control?

References

1. https://www.annualreviews.org/doi/10.1146/annurev-statistics-040220-091526
2. https://www.sciencedirect.com/topics/economics-econometrics-and-finance/diffusion-of-innovation
3. https://arxiv.org/pdf/0706.0641
4. https://link.springer.com/article/10.1186/s40064-016-1675-x
5. https://ideas.repec.org/a/eee/respol/v29y2000i4-5p603-625.html
6. https://www.sciencedirect.com/science/article/abs/pii/S0167718705000081
7. https://www.academia.edu/26656163/Financial_Factors_and_the_Inter_Firm_Diffusion_of_New_Technology_A_Real_Options_Model
8. https://www.sciencedirect.com/science/article/abs/pii/016771878990012X
9. https://repub.eur.nl/pub/827/rm0009.pdf
10. https://research.vu.nl/en/publications/a-vintage-of-model-technology-diffusion-the-effect-returns-to-div
11. https://www.researchgate.net/publication/226649759_Technology_Diffusion_in_Energy-Economy_Models_The_Case_of_Danish_Vintage_Models