Who Is Responsible for Climate Change? A Four-Lens Analysis
Introduction
Climate change is one of the most consequential policy challenges of the 21st century, yet the question of who bears responsibility for it remains deeply contested. The answer matters enormously: it determines which countries must cut emissions fastest, which receive climate finance from wealthier nations, and which face legal and moral liability for losses already suffered by vulnerable populations. Billions of people — particularly those in low-income countries who have contributed least to the problem — will live with the consequences of whatever agreements are reached. Getting the measurement right, or at least understanding how measurement shapes the debate, is not an academic exercise. It is a prerequisite for informed climate citizenship. When world leaders gather at climate summits, they rarely agree on who is most responsible — and a large part of that disagreement comes down to measurement. Depending on which metric a country chooses to cite, it can appear to be either a major culprit or a relatively minor contributor. China points to its low per-capita emissions. The United States defends its recent reductions in annual output. Developing nations invoke historical accumulation. Each argument is backed by real data. Each leads to a different verdict.
This project asks a deceptively simple question: who is responsible for climate change? Rather than providing a single answer, we argue that the question cannot be answered without first specifying what kind of fairness we believe should govern climate responsibility. We examine four empirically valid metrics: total annual emissions, per-capita emissions, cumulative historical emissions, and consumption-based emissions, applied to seven major emitters: China, the United States, India, Germany, the United Kingdom, Russia, and Japan. Our goal is not to declare a winner, but to make the terms of the debate visible. This story is designed for anyone trying to read a climate headline, a UN negotiation brief, or a political speech more critically. By presenting all four lenses in a structured scrollytelling format, we invite readers to see how the same underlying data can tell four very different stories — and to recognize when a particular framing is being used to serve a particular interest.
Data
Our analysis draws on the Our World in Data CO₂ and Greenhouse Gas Emissions dataset, 2024 release. The dataset is a country-year panel covering emissions records from as early as 1750, with r nrow(readr::read_csv("data/raw/owid-co2-data.csv", show_col_types = FALSE)) rows and r ncol(readr::read_csv("data/raw/owid-co2-data.csv", show_col_types = FALSE)) columns. Key variables include total annual CO₂ emissions (co2), per-capita emissions (co2_per_capita), cumulative historical emissions (cumulative_co2), and consumption-based emissions (consumption_co2).
We focused our analysis on seven countries selected to represent the key fault lines in the climate responsibility debate: China and India as large, rapidly industrializing economies; the United States, Germany, and the United Kingdom as early industrializers and high-income importers; Russia as a major fossil fuel producer; and Japan as a high-income Asian economy with a distinct energy profile.
Data cleaning involved removing aggregate regional entries (e.g., “World”, “Asia”) that lack ISO country codes, selecting only the columns relevant to our four metrics, and filtering for years from 1900 onward for time-series analyses. Consumption-based emissions data is only available from 1990 onward, which limits the scope of Lens 4 relative to the others. All cleaned datasets are stored in data/processed/.
Design Process (Justification Approach)
Why Scrollytelling?
Our central design decision was to present this analysis as a Closeread scrollytelling narrative rather than a static report or interactive dashboard. The core challenge of this topic is not a lack of data — it is that readers encounter these metrics in isolation, without understanding how they relate to one another or what assumptions each one encodes. A scrollytelling format allows us to guide readers through each lens sequentially, building understanding one step at a time before asking them to hold all four perspectives simultaneously.
We chose Closeread specifically because it allows the sticky visualization panel to remain in view while the narrative scrolls, enabling readers to follow the argument and observe the data at the same time. This is particularly effective for our use case, where each section introduces a new framing of the same underlying data.
Narrative Structure
We organized the story into four chapters, each corresponding to one lens. The order was deliberately chosen to move from the most familiar and politically convenient measure (total output) to progressively more challenging framings (per capita, cumulative, consumption). This progression is intentional: by the time readers reach Lens 4, they have enough context to appreciate why the consumption framing is so politically charged for wealthy nations.
Within each chapter, we structured the narrative into three beats: an introduction to the metric and what it measures, a close reading of what the data shows, and a reflection on the limitations and political implications of that lens. This three-beat structure creates a consistent rhythm that makes the story easy to follow across all four chapters.
Visual Design
We chose a restrained, editorial visual language — drawing on the aesthetic of long-form data journalism — to signal to readers that this is a serious analytical piece rather than an advocacy document. The color palette (#F2F0EB background, #1a1a1a text and accents) is neutral and high-contrast. Typography combines Playfair Display for headings (authoritative, editorial), Lora for body text (readable, literary), and Inter for labels and UI elements (clean, functional).
Country colors were assigned consistently across all four plots to enable readers to track individual countries across charts. China (#E63946) and the United States (#1D3557) were assigned the most visually distinct colors, as they are the two countries whose relative positions change most dramatically across the four lenses.
Interactivity
We added two interactive elements to improve reader engagement. First, the opening section uses a card-reveal mechanism: as readers scroll through the introduction, each of the four lens cards appears progressively, giving readers a preview of the story structure before it begins. Second, the conclusion section uses an animated bar fill effect, where each verdict bar fills from left to right with a staggered delay, drawing the reader’s attention sequentially through the four findings.
Alternatives Considered
We considered building a Shiny dashboard that would allow users to select metrics and countries interactively. We decided against this approach because interactivity without narrative guidance can make it harder, not easier, for readers to understand the significance of what they are seeing. The scrollytelling format imposes a reading order that serves the argument. We also considered including more countries, but limiting to seven allowed us to maintain visual clarity in the line charts without overcrowding.
Key Findings
Lens 1 — Total Emissions:
By annual output, China is the world’s largest emitter by a significant margin, having surpassed the United States around 2006. However, this metric reflects the scale of an economy rather than the carbon intensity of individual lives. China’s rise is largely attributable to its role as the world’s manufacturing hub, producing goods consumed by wealthier nations.
Lens 2 — Per Capita Emissions:
When emissions are divided by population, the ranking changes substantially. The United States, Russia, and Germany have historically been among the highest per-capita emitters, reflecting car-dependent infrastructure, energy-intensive industry, and fossil-fuel-dependent power grids. India and China, despite their large aggregate output, emit considerably less per person — a reflection of the hundreds of millions of citizens who still have limited access to electricity and private vehicles.
Lens 3 — Cumulative Emissions:
Accounting for the longevity of CO₂ in the atmosphere, the United States has contributed approximately 23.5% of all historical emissions since industrialization began — more than any other country. The United Kingdom and Germany, though modest current emitters, also carry a disproportionate historical burden as early industrializers. China’s cumulative total, despite its high current output, is less than half that of the United States, because its large-scale industrialization only accelerated after 1980.
Lens 4 — Consumption vs. Production:
Standard emissions accounting assigns CO₂ to the country where goods are produced. Consumption-based accounting reassigns it to where those goods are consumed. Under this framework, the United States and United Kingdom appear substantially more responsible than production figures suggest. Their apparent domestic progress — falling annual emissions since the 1990s — is partly an artifact of offshoring manufacturing to countries like China, rather than genuine decarbonization.
Limitations
Several limitations affect the scope and interpretation of our findings. Importantly, we tried to anticipate each one during the design process rather than simply note it after the fact.
Data coverage: Consumption-based emissions data is available only from 1990 onward and is missing for many country-years, particularly smaller economies not captured by global trade flow models. This constrains Lens 4 to a shorter and less complete time window than the other three lenses. We addressed this limitation directly in the design: the Lens 4 narrative explicitly flags the 1990 cutoff, and we avoided any claims that would require pre-1990 consumption data. The chart itself is scoped to the period where data is reliable.
Historical data quality: Cumulative emissions estimates before 1850 carry greater uncertainty than modern figures, since early industrial activity was not systematically recorded. This means our cumulative totals for early industrializers like the United Kingdom may slightly understate their true historical contribution. In the narrative, we deliberately frame these figures as conservative lower bounds rather than precise measurements, and we avoided language that would overstate the precision of historical estimates.
Country selection: Limiting to seven countries was a conscious design tradeoff. A more comprehensive analysis would include a wider range of emitters — particularly smaller developing nations that contribute minimally to emissions but face the greatest climate impacts. We prioritized visual clarity in the line charts over comprehensiveness, and we acknowledge in the narrative that the seven countries were chosen to represent the key fault lines in the responsibility debate, not to provide a complete global picture.
Scope of emissions: Our analysis covers CO₂ only, excluding methane, nitrous oxide, and other greenhouse gases that would shift some rankings — particularly for countries with large agricultural sectors. We chose CO₂ because it is the dominant driver of long-run warming and has the most complete historical record. Future iterations of this project could extend the analysis to a full greenhouse gas basket using the ghg variables available in the same OWID dataset.
Metric limitations: No single lens is sufficient on its own, and there is no agreed-upon method for combining the four into a composite score. Rather than proposing one — which would require normative choices the data cannot support — we designed the conclusion to present the lenses as complementary tools for critical reading. The summary table and verdict visualization in the conclusion were specifically built to hold all four perspectives simultaneously without collapsing them into a single ranking.
Conclusion
The question of who is responsible for climate change does not have a single correct answer — but it does have several honest ones, each grounded in a different understanding of fairness. Total emissions place the burden on those who currently pollute the most. Per-capita emissions hold every person to the same standard regardless of nationality. Cumulative emissions hold early industrializers accountable for the long tail of their historical output. Consumption-based emissions shift responsibility from where goods are made to where they are wanted.
By presenting all four lenses in a structured narrative, this project aims to equip readers with the analytical tools to evaluate climate responsibility claims critically — to ask not just “which country emits the most?” but “which metric is being used, and whose interests does it serve?” In a policy environment where data is routinely deployed selectively to support predetermined positions, that kind of critical literacy is, we believe, a meaningful contribution.
The data cannot tell us what is fair. But it can tell us who has benefited most from the way responsibility has historically been measured — and who has not.