🔬 physics

An Analytics Framework for Modeling Residential Photovoltaic Adoption and Decision Dynamics

This study presents a data-driven analytics framework that models residential photovoltaic adoption in Catalonia using a logistic growth function and spatial analysis, revealing that imitation effects and social perception are the primary drivers of adoption decisions over regulatory and socioeconomic factors.

Original authors: Canigó Callau-Boix, Raúl Toral, Pere Colet

Published 2026-05-22

📖 5 min read🧠 Deep dive

CC BY 4.0

Original authors: Canigó Callau-Boix, Raúl Toral, Pere Colet

Original paper licensed under CC BY 4.0 (http://creativecommons.org/licenses/by/4.0/). ✨ This is an AI-generated explanation of the paper below. It is not written or endorsed by the authors. For technical accuracy, refer to the original paper. Read full disclaimer

Imagine the spread of rooftop solar panels in Catalonia (a region in Spain) not as a complex engineering problem, but as a giant game of "copycat" played by thousands of neighbors. This paper is a detailed report on how that game was played, who played it, and what rules changed the game's outcome.

Here is the story of the paper, broken down into simple concepts:

1. The Main Idea: The "Copycat" Effect

The researchers wanted to understand why people decide to put solar panels on their roofs. They looked at data from 2002 to 2024, covering over 112,000 installations.

They found that the most powerful force driving this decision wasn't a government ad campaign or a sudden drop in the price of panels. It was imitation.

The Analogy: Think of a new dance move. At first, only a few people know it. But once a few neighbors start doing it, others see it and think, "Oh, that looks cool, I want to do that too." The more people who do it, the more others want to join in.
The Finding: The data showed that "innovation" (people deciding to do it just because they read about it in a magazine) was almost zero. The entire growth was driven by "contagion"—seeing your neighbor do it and deciding to do it yourself.

2. The Two Acts of the Play

The history of solar adoption in Catalonia didn't happen in a straight line; it happened in two distinct "acts," like a play with an intermission.

Act 1 (The Slow Start): From 2002 to 2015, adoption was slow and then hit a wall. Why? Because of a confusing tax called the "Solar Tax" (impuesto al sol). Even though this tax didn't technically apply to small home systems, the public was confused and scared. People thought, "If I put panels on my roof, I might get fined." So, everyone stopped. The play went silent.
The Intermission (2015–2016): The tax was repealed, and the confusion cleared up.
Act 2 (The Boom): Starting in 2016, the "copycat" effect kicked into high gear. Adoption exploded. The researchers predict this boom is now reaching its "end of the line" (saturation), meaning most people who can easily install panels have already done so.

3. The "Social Temperature" vs. The "Real Price"

The researchers tried to figure out what made the "copycat" effect speed up or slow down. They looked at two things:

The Real Price of Electricity: How much it actually cost to buy power from the grid.
The "Social Temperature": How much people were worried about the price (measured by how many people Googled "electricity price").

The Surprise: The actual price of electricity mattered less than the worry about it.

The Analogy: Imagine a fire alarm. The actual fire (high prices) is dangerous, but the sound of the alarm (people talking about high prices and Googling it) is what makes everyone run out the door. The study found that when people were talking and searching about high prices, they were much more likely to install solar panels, even if the actual price hadn't changed much yet.

4. The Map: Who Gets the Panels?

The researchers looked at a map of Catalonia to see where the panels were going.

Population is Key: The number of panels in a town generally matched the number of people living there. It's like a shadow following a person; where the people are, the panels tend to be.
The "Apartment Problem": However, the map revealed a big gap. The places with the fewest panels per person were the crowded cities with tall apartment buildings.
- Why? It's hard to put solar panels on a roof shared by 50 families. You need permission from everyone, and the roof might be shaded by other buildings. It's like trying to plant a garden in a shared balcony where everyone has to agree on the soil.
- The Result: People in small towns or houses with their own roofs (low-rise buildings) adopted solar much faster than people in high-rise cities.

5. The Conclusion: We Are Almost Full, But There's Room to Grow

The study concludes that the "easy" phase of solar adoption is over. The growth curve is flattening out, like a balloon that has been blown up to its limit.

However, the balloon isn't fully inflated yet. The researchers estimate that only about 12% of all homes in the region have panels. The reason we aren't at 100% isn't because the sun stopped shining or the technology broke; it's because of social and legal hurdles, especially for people living in apartments.

The Final Takeaway:
To get more solar panels, we don't just need cheaper technology. We need to:

Keep the "social temperature" high (keep people talking about energy costs).
Fix the "Apartment Problem" by making it easier for neighbors in big buildings to agree on shared solar projects.

The paper suggests that if we clear these hurdles, the "copycat" game can start a new round, especially in the crowded cities where the game hasn't really started yet.

Technical Summary: An Analytics Framework for Modeling Residential Photovoltaic Adoption and Decision Dynamics

Problem Statement
Photovoltaic (PV) generation is central to the global energy transition, yet understanding the specific dynamics of residential self-consumption adoption requires robust analytical frameworks that capture both temporal and spatial decision patterns. While existing literature identifies economic, environmental, and regulatory factors as drivers of adoption, the specific role of social interactions and the distinction between large-scale commercial plants and residential systems often remain under-analyzed. Furthermore, many studies rely on limited datasets or fail to distinguish between "innovation" (external influence) and "imitation" (social influence) in driving adoption curves. This study addresses the need for a data-driven framework to model the diffusion of residential PV in Catalonia, Spain, spanning from 2002 to 2024, to better inform policy and investment decisions.

Methodology
The study utilizes a comprehensive dataset from the Registre d'Autoconsum a Catalunya (RAC), comprising 112,086 geolocated residential PV installations (defined as systems <10 kW) up to May 31, 2024. The methodology integrates three primary analytical components:

Temporal Modeling: The adoption process is modeled using the Bass model and the simpler Logistic Model (LM). The authors fit these models to the cumulative number of installations to determine the relative contributions of innovation (external factors, parameter $p$ ) and imitation (social influence, parameter $q$ ). Model selection is performed using the Bayesian Information Criterion (BIC).
External Factor Analysis: To quantify the influence of external variables (subsidies, tax deductions, electricity prices) without imposing a priori functional forms, the authors employ a generalized Bass model approach. They empirically infer a time-dependent scaling function, $\chi(t)$ , which represents deviations from the baseline logistic growth. This function is then correlated with external data, including government subsidy periods, tax deduction policies, electricity prices (PVPC), and Google search volumes for terms like "precio luz" (electricity price) and "impuesto al sol" (solar tax).
Spatial Analysis: The study conducts a spatial analysis at the municipal and comarca (regional) levels. It examines correlations between installation density and demographic/socioeconomic variables (population, income, building height) and analyzes spatiotemporal correlations using pair distribution functions ( $C(\tau)$ for time and $\rho(r)$ for space) to assess clustering and diffusion patterns.

Key Results

Dominance of Imitation: The logistic model provides a significantly better fit (lower BIC) than the Bass model for the main growth phase (June 2016–May 2024). The innovation coefficient ( $p$ ) in the Bass model is statistically consistent with zero, indicating that adoption is driven primarily by imitation (social influence) rather than external innovation drivers alone.
Growth Phases and Saturation: The analysis identifies two distinct logistic growth phases. The first phase (2013–2015) reached a low saturation level, followed by a stagnation period. The second phase began in June 2016, reaching an inflection point in early 2023 and approaching a saturation capacity of approximately 118,600 installations. The current adoption level (112,086 by May 2024) suggests the market is nearing saturation.
Role of Social Perception: The empirical scaling function $\chi(t)$ reveals that while direct subsidies (Catalan government) and electricity price increases correlate with adoption spikes, the social perception of electricity prices (measured by Google search volume) exhibits a stronger and more immediate correlation with adoption dynamics than the actual price itself. Conversely, the "solar tax" (impuesto al sol), despite applying only to systems >10 kW, caused a widespread stagnation in residential (<10 kW) adoption due to public confusion and misinformation.
Spatial Heterogeneity: Residential adoption follows a power-law distribution relative to population ( $R^2 = 0.8$ , slope $\approx 0.84$ ), indicating that adoption scales with population density but with a saturation effect in highly populated areas. Crucially, adoption per capita is negatively correlated with urban population density and positively correlated with the fraction of buildings under three stories. High-density urban areas with multi-apartment buildings show significantly lower adoption rates due to roof space constraints, shared ownership complexities, and regulatory hurdles.
Spatiotemporal Correlations: The temporal pair distribution of installations is well-described by the analytical expression derived from logistic growth, suggesting a homogeneous diffusion process. Spatially, the distribution of installations closely mirrors the distribution of residential buildings, indicating that the built environment's structure, rather than specific short-range social clustering, dictates spatial adoption patterns.

Key Contributions

Methodological Innovation: The paper proposes an original method to empirically estimate the influence of external factors on adoption dynamics by inferring the scaling function $\chi(t)$ directly from data, avoiding assumptions about the functional form of these influences.
Differentiation of Drivers: It provides evidence that for residential self-consumption, imitation effects are the primary driver of adoption, rendering the innovation parameter negligible in the Bass model context for this specific sector.
Policy Insight: The study highlights that social perception (e.g., public concern over electricity prices) can be a more potent driver of adoption than the economic variables themselves, and that regulatory uncertainty (even regarding taxes not directly applicable to a sector) can cause significant market stagnation.
Spatial Diagnostics: It identifies specific structural barriers to adoption in dense urban environments (multi-story buildings), suggesting that future growth potential lies in overcoming collective installation barriers rather than just individual incentives.

Significance and Claims
The authors claim that their framework offers a structured approach for predictive and diagnostic analytics in energy transition planning. The findings suggest that the slowdown in residential PV adoption in Catalonia is not due to physical constraints (as the saturation level represents only ~12% of total residential buildings) but rather to regulatory and socio-economic contexts. The paper asserts that future expansion is possible if legislative changes are introduced and, critically, if they are clearly perceived by the public. The study concludes that while the models used are simplified (ignoring specific individual interactions), they effectively capture the macro-dynamics, likely due to the small-world properties of social networks and the relevance of weak ties in information diffusion. The framework is presented as applicable to other regions provided detailed geolocated data is available.

1. The Main Idea: The "Copycat" Effect

2. The Two Acts of the Play

3. The "Social Temperature" vs. The "Real Price"

4. The Map: Who Gets the Panels?

5. The Conclusion: We Are Almost Full, But There's Room to Grow

More like this