{ "schema": "ev-reading-guides/v1", "revision": "r6", "sourceEdition": "2026-09-18", "storyPath": "/blog/to-charge-or-not-to-charge", "reader": { "title": "What can an electric-car curve actually tell us?", "deck": "This is an investigation of adoption, prediction and fleet replacement. Those are three different problems, although public debate often compresses them into a single percentage.", "questions": [ { "number": "01", "title": "How far has the market moved?", "answer": "A share of new registrations measures the current purchase mix. A share of the active fleet measures the accumulated stock. Compare both before deciding whether the transition looks fast or slow.", "href": "#two-clocks", "link": "Read the two clocks" }, { "number": "02", "title": "Can a curve predict the next turn?", "answer": "A logistic or Gompertz curve is a mathematical model, not an observed law of adoption. Its forecast has to earn its place against a benchmark at a stated horizon, using a stated sample.", "href": "#model-guide", "link": "Understand the curves" }, { "number": "03", "title": "When does the existing fleet change?", "answer": "The fleet depends on additions, exits and other net adjustments over time. Even a fully electric new-car market leaves older vehicles on the road. A scenario can expose that mechanism without predicting the future.", "href": "#long-road", "link": "Follow the stock arithmetic" } ], "takeaway": "The transition is observable; a universal timetable is not. S-curves can be useful over some horizons, but countries, policy interruptions and the stock-flow distinction prevent one attractive line from settling every decision.", "scope": "Unless explicitly stated otherwise, electric means battery-electric (BEV), not BEV plus plug-in hybrid. Research edition: 18 September 2026. The site retains the supplied econometric estimates; it does not claim to have refitted the models.", "dataTitle": "Know what each dataset can answer.", "dataDeck": "The evidence is strongest when the question matches the measurement. The passports below identify the population, period and limitation before the charts ask you to compare them.", "datasets": [ { "id": "international", "label": "International market histories", "source": "IEA / Our World in Data", "coverage": "60 countries; 759 national country-year observations, 2010–2025. Fifty-nine countries report 2025; 48 enter the forecast contest.", "measure": "BEVs as a percentage of new car sales. The IEA Cars definition includes SUVs and passenger pickups; it is not a uniform European M1 registration census.", "question": "How do national paths, threshold crossings and forecast errors differ?", "boundary": "Shares are rounded and implied counts approximate. A country's first reported crossing is an annual observation, not the exact date of a causal tipping point.", "href": "/data/ev-diffusion/tables/global_country_year.csv" }, { "id": "registrations", "label": "Germany's registration flow", "source": "KBA", "coverage": "116 monthly observations, January 2017–August 2026.", "measure": "New BEV passenger-car registrations divided by all new passenger-car registrations in the same period. A trailing year uses the ratio of twelve monthly sums.", "question": "How quickly is the mix of newly registered cars changing?", "boundary": "A new registration is not necessarily a household's first electric car. The 2026 record ends in August; it is not a completed calendar year.", "href": "/data/ev-diffusion/tables/germany_monthly.csv" }, { "id": "fleet", "label": "Germany's active fleet", "source": "KBA", "coverage": "Six supplied snapshots: January 2022, 2023, 2024, 2025 and 2026, plus 1 July 2026.", "measure": "Registered BEV passenger cars divided by all registered passenger cars on each stated reference date.", "question": "How much of the accumulated vehicle stock has changed?", "boundary": "The fleet is not observed monthly in this package. Reference and publication dates differ, and a missing earlier snapshot is not a zero.", "href": "/data/ev-diffusion/tables/germany_stock.csv" }, { "id": "districts", "label": "Registered-address geography", "source": "KBA / attributed district boundaries", "coverage": "400 mapped registration districts in the January 2026 atlas. A separate July 2026 stock snapshot is retained in the source package.", "measure": "BEV share within each district's registered passenger-car fleet. National totals also include an unmapped unallocated category.", "question": "Where are BEVs registered, and which regional associations merit closer investigation?", "boundary": "A keeper's address is not necessarily where a car is driven or charged. Corporate fleets can influence the pattern; contemporary boundaries are not a fixed historical geography.", "href": "/data/ev-diffusion/tables/germany_region_year.csv" }, { "id": "charging", "label": "Charging and housing context", "source": "Bundesnetzagentur / Census 2022", "coverage": "Matched district snapshots; the displayed association compares levels with 2023–2026 changes. Housing characteristics refer to the 2022 census.", "measure": "Public charging-point counts and regional housing characteristics, used as contextual variables in the supplied analyses.", "question": "How sensitive is an infrastructure association to the specification and to changes over time?", "boundary": "Points are not sites or connectors. Counts do not measure charging utilization or an individual household's access. Inconsistent source totals remain flagged or missing, not repaired by guesswork.", "href": "/data/ev-diffusion/tables/regional_associations.csv" }, { "id": "models", "label": "Models, diagnostics and scenarios", "source": "Supplied research calculations", "coverage": "Country backtests, German forecast origins, bootstrap comparisons, saturation profiles and 2026–2040 fleet scenarios.", "measure": "Predictions and their errors, parameter-profile loss, conditional associations, and fixed-assumption stock arithmetic. Each has a different interpretation.", "question": "What remains credible after a model is compared with observations and its assumptions are varied?", "boundary": "These are not additional observed cars. The model outputs were supplied with the research; neither an uncertainty interval nor an assumption range automatically predicts an electrification date.", "href": "#methods" } ], "evidenceKinds": [ { "label": "Observed", "meaning": "A reported administrative or statistical observation, subject to its stated scope and revisions." }, { "label": "Derived", "meaning": "Arithmetic applied to observations: a share, change, weighted average or accounting decomposition." }, { "label": "Modelled", "meaning": "A fitted relationship, forecast, error comparison or diagnostic from the supplied research." }, { "label": "Scenario", "meaning": "What follows if the explicitly chosen future assumptions hold. Not a prediction interval." }, { "label": "Illustrative", "meaning": "A teaching curve, schematic image or reader sketch. It is not an empirical estimate." } ], "implications": [ { "audience": "Manufacturers and suppliers", "title": "Separate the next planning horizon from the terminal share.", "text": "Use a forecast only where it improves the relevant benchmark, then stress it against plausible interruptions. A fitted ceiling does not validate a factory timetable, and a one-year advantage does not establish long-run certainty." }, { "audience": "Charging operators and planners", "title": "Plan around use and access, not only registered addresses.", "text": "The map identifies a registered fleet, not charger occupancy. Combine it with travel patterns, parking, private charging access and utilization evidence before translating a regional correlation into capacity or site decisions." }, { "audience": "Policymakers", "title": "Report progress in both the flow and the stock.", "text": "A strong registration month is not a complete fleet transition. Distinguish announcement, eligibility, applications and observed purchases, and do not infer a causal subsidy effect from chronology alone. These data also do not measure lifetime emissions or mileage." }, { "audience": "Readers and households", "title": "Use the evidence to understand the market, not to outsource a purchase decision.", "text": "The dataset cannot supply your transaction price, parking arrangement, journeys or charging tariff. Those belong in a separate total-cost and practical-suitability calculation. A national adoption curve is context, not a recommendation to buy a particular car." } ] }, "figures": { "ev-01": { "question": "Why can the transition look fast in the showroom and slow on the road?", "read": "The horizontal axis is calendar time; the vertical axis is a BEV percentage. The registration line is monthly. The separately dated fleet marks are snapshots, not a monthly stock series.", "definition": "Registration share divides new BEVs by all new passenger cars during a period. Fleet share divides registered BEVs by all registered passenger cars on a date. Their denominators and clocks differ.", "meaning": "The gap shows why a change in the purchase mix reaches the accumulated fleet gradually. A volatile month need not describe the trailing year or the cars already in use.", "decision": "Use the registration clock for near-term market monitoring and the fleet clock for the accumulated transition. Neither is a direct measure of first-time household adoption.", "caution": "The reference-date join uses this revised research edition, not information necessarily published at each historical month end." }, "ev-02": { "question": "Are countries following one common path at different dates?", "read": "Each line follows a country's BEV share of new car sales on the same zero–100% scale. Start with the selected market and the labeled comparison; use the controls to inspect other countries or widen the context.", "definition": "These are annual IEA/OWID car-market shares, not German KBA registrations or active-fleet shares. An endpoint keeps its actual last observed year.", "meaning": "The differences include both levels and changes in direction. A market at a low share is not automatically a delayed copy of Norway, and a high share does not make the preceding path smooth.", "decision": "A forecast or capacity plan needs evidence from the market and horizon it will serve. Do not substitute an unweighted visual average of country curves for a global market-weighted share.", "caution": "The reference markets are reading aids, not a representative sample. The full record contains 60 countries and remains available." }, "ev-03": { "question": "Does crossing five percent start a reliable countdown?", "read": "Time zero is each country's first supplied annual observation at or above 5%. Moving right follows later years for that country; it does not move every country through the same calendar years.", "definition": "The starting value can exceed 5% because observations are annual. Countries that never cross in their available record remain censored; they are not assigned a zero-length successful transition.", "meaning": "The alignment removes one difference in timing but not differences in institutions, product offerings, policy or market structure. Paths still diverge and sometimes reverse.", "decision": "Use the milestone to organize a comparison, then test the mechanisms and uncertainties that matter in the actual market. The threshold itself does not supply a causal explanation or arrival date.", "caution": "Available follow-up differs by crossing year. A later entrant cannot be given an invented five- or ten-year future." }, "ev-04": { "question": "How much does today's market share tell us about next year's gain?", "read": "Read a point across, then up. Its horizontal coordinate is the BEV share in year t. Its vertical coordinate is the change in the following year, measured in percentage points. The zero line separates gains from declines.", "definition": "A move from 5% to 10% is a gain of five percentage points, even though the share doubles. Only consecutive annual observations form a point in this comparison.", "meaning": "Different subsequent changes occur at similar starting shares. The selected connected path supplies its country and time order; the background points show that current penetration alone does not pin down the next observation.", "decision": "Stress a smooth adoption narrative against pauses and reversals rather than treating the current level as a sufficient forecasting variable.", "caution": "This comparison challenges a universal common-parameter path; it does not identify which intervention caused a country's change." }, "ev-05": { "question": "Does the model forecast better, or does it merely fit the past attractively?", "read": "A filled dot is a model's mean absolute error; the open dot is the naive benchmark on the same horizon. Further left is better. The connector compares errors, not two dates. Switch the horizon to repeat the contest two years ahead.", "definition": "Absolute error is the distance between a predicted and an observed share in percentage points. The study averages within a country and then gives eligible countries equal weight. A large market does not get a larger vote.", "meaning": "At one year the Gompertz error is 2.65 points against 2.94 for naive. That is a modest average advantage, not a guarantee that the curve wins for every country, origin or decision.", "decision": "Require an adoption model to beat a transparent benchmark out of sample before using it to plan capacity. Read the uncertainty comparison next; a point ranking is only the beginning of that test.", "caution": "The 48 eligible countries have sufficient consecutive histories. This is a revised-history exercise, not a reconstruction of everything forecasters knew at the time." }, "ev-06": { "question": "How secure is the apparent forecasting advantage?", "read": "The vertical zero line means the same error as naive. Dots to its left favor the fitted model; dots to its right favor naive. A horizontal whisker gives a supplied 95% bootstrap interval for that error difference.", "definition": "Country clustering keeps a country's repeated forecast errors together. Origin blocking instead respects common shocks at forecast dates. These procedures also change weighting, so their centers need not coincide.", "meaning": "An interval crossing zero leaves both an advantage and a disadvantage compatible with that diagnostic. With only nine one-year forecast origins, the origin-block evidence is especially limited.", "decision": "Distinguish an average ranking from a robust planning edge. Ask whether the result survives the dependence structure relevant to the decisions, rather than quoting the narrowest whisker.", "caution": "These are intervals for an average error comparison, not prediction bands around future BEV sales. They cannot be transferred to the next chart as a forecast fan." }, "ev-07": { "question": "What happens when a fitted path meets observations it has not seen?", "read": "The solid line is the subsequently observed monthly BEV registration share. Dashed lines are the three displayed model predictions, frozen at the selected December origin. The marked origin separates the information cutoff from later target months.", "definition": "The forecasts are computed once at the origin using the supplied revised history; later observations do not update them. Naive repeats the previous annual seasonal cycle, while the structural models extrapolate their fitted shape.", "meaning": "This test exposes the difference between a smooth underlying path and a monthly market that can move abruptly. It is useful precisely because the observed record is allowed to disagree with the model.", "decision": "Treat a model as a conditional planning instrument and monitor departures from it. A short-horizon comparison cannot certify the date when all new cars, still less the active fleet, become electric.", "caution": "The cutoff is a retrospective fitting origin, not a vintage-correct archive. Policy timing is context; proximity to a jump does not establish a causal effect." }, "ev-08": { "question": "How much does China's market weight change the global reading?", "read": "Compare the world line with China and with the world excluding China. They share a percentage scale, but they are overlapping aggregates—not three categories that can be added.", "definition": "The global share weights markets by their inferred new-car volumes. It is not the simple average of national percentages. Ex-China is recalculated from the remaining approximate counts.", "meaning": "In 2025 the world BEV share is 16.51%, while the result excluding China is 9.71%. Market size therefore changes what the global headline represents.", "decision": "Use a global figure for global exposure only after inspecting its composition. A supplier focused elsewhere should not read the world average as its own addressable-market forecast.", "caution": "This is an accounting exclusion, not a causal estimate of what would happen if China changed its behavior." }, "ev-09": { "question": "Did the world share rise because markets electrified or because their weights changed?", "read": "Follow the contributions from left to right. The components reconcile to the 2024–2025 increase in the global BEV share; the unit is percentage points, not a count of vehicles.", "definition": "The two-factor decomposition separates changes within markets from changes in their relative market weights. The reviewed calculation uses a balanced 53-country panel and an explicit residual.", "meaning": "Rising within-market penetration accounts for about 88% of the 3.31-point global increase. The remaining contribution comes from the change in market composition.", "decision": "The result distinguishes two sources of an aggregate change. It prevents a story about market size from erasing the simultaneous changes inside markets.", "caution": "Accounting contributions are not causal effects. The exercise does not recreate a world in which market sizes or national behavior were held fixed by an intervention." }, "ev-10": { "question": "Whose registrations account for Germany's stronger electric mix?", "read": "Each line divides BEVs by all registrations within the same keeper category. Compare the private and commercial rates, rather than adding their percentages.", "definition": "A registered keeper is an administrative category. Commercial registrations include dealers, manufacturers and rental firms; private registrations do not identify first-time adoption or the household's prior fleet.", "meaning": "Across January–August 2026, the BEV share is 34.77% among private keepers and 21.56% among commercial keepers. The aggregate rebound therefore has an important private-registration component.", "decision": "Market planning should distinguish these channels and their purchasing calendars. One national average can conceal a materially different mix of buyers and registered keepers.", "caution": "Category differences and policy dates do not isolate a subsidy effect. The year-to-date window is not a full 2026 result." }, "ev-11": { "question": "Where are BEVs registered, and what can the map not locate?", "read": "The shading is a rate: registered BEVs divided by all registered passenger cars in each district. Focus or select an area for counts, its exact share and a rank among the 400 mapped districts.", "definition": "The map uses the January 2026 registered-address snapshot and attributed contemporary boundaries. It is not a map of journeys, overnight parking, public-charger use or individual purchasing motives.", "meaning": "The displayed district rates range from about 1.24% to 12.50%. Corporate fleets and registered addresses can help explain why a place looks different from the national fleet without measuring household preferences directly.", "decision": "Use an unusual district as a question for further investigation. Combine the register with use, travel and access evidence before making a location decision.", "caution": "The national unallocated category is not mapped. Equal rates receive equal ranks; a shaded boundary is not a causal finding." }, "ev-12": { "question": "Does more public charging identify a stronger electric-car transition?", "read": "The point is a charging coefficient from a stated regression specification. Its whisker is an approximate 95% interval. The vertical reference at zero marks no estimated conditional association on this scale.", "definition": "The dependent quantity is BEV fleet share in percentage points. The charging variable is log(1 + charging points), which compresses large counts. The coefficient is not an effect per additional charger.", "meaning": "The positive association in levels is not retained in the displayed 2023–2026 change specification. The answer is sensitive to what is being compared, not merely to how the chart is styled.", "decision": "Treat infrastructure and vehicle adoption as jointly determined planning questions. These estimates do not identify the return from installing a new point in a district.", "caution": "The intervals use state-cluster standard errors with only 16 clusters and 395 matched areas. They do not remove omitted variables or reverse causality." }, "ev-13": { "question": "How much of the ceiling is learned from the observations?", "read": "The horizontal axis is an imposed long-run sales-share ceiling K, not a year. For each ceiling, the remaining logistic parameters are refitted. The vertical coordinate is the resulting in-sample root-mean-square error in percentage points.", "definition": "Root-mean-square error squares the fitting errors, averages them and takes the square root. This profile uses 60 fixed ceilings from max(5%, 90% of the observed maximum) to 100%, refitting the other parameters at each point. Its minimum is a grid minimum, not a certain destination.", "meaning": "If a wide range of ceilings gives nearly the same loss, the available history does not identify the ceiling tightly. An optimizer may return a precise number even when the economic conclusion is weak.", "decision": "Inspect the range of plausible model shapes and the observation window before promising an adoption date. A shallow loss profile calls for sensitivity analysis and additional evidence, not more decimal places.", "caution": "This is a diagnostic loss curve, not a probability distribution. Both axes adapt to the selected country's tested ceilings and losses; do not compare countries by the apparent steepness of differently scaled plots." }, "ev-14": { "question": "Is a young electric fleet the same thing as an electric young-car market?", "read": "Each bar shows BEVs as a share of all cars in one age group. Its denominator changes from group to group; the bars are not the age composition of BEVs.", "definition": "BEV penetration among cars younger than five divides young BEVs by all young cars. The statement that 89.25% of BEVs are younger than five divides young BEVs by all BEVs instead.", "meaning": "BEV shares are 15.46% among cars younger than five and 1.47% among five-to-nine-year-old cars. The contrast is consistent with a recent expansion entering an older accumulated fleet.", "decision": "A strong new-car mix does not erase older cohorts. That matters when interpreting the speed at which sales changes can alter the active stock.", "caution": "One age cross-section does not estimate how long a car will survive or the probability that a BEV leaves the register." }, "ev-15": { "question": "Why did 545,142 new BEVs not add the same number to the active fleet?", "read": "Follow the signed accounting components. New registrations plus the net adjustment reconcile to the observed increase between the January 2025 and January 2026 stock snapshots.", "definition": "The stock increase is 382,617 cars. Subtracting new BEV registrations produces a residual of −162,525. That is an observed reconciliation, not an independently measured flow of exports.", "meaning": "The register is affected by more than new registrations. The residual combines unobserved used imports, exports, scrappage and register adjustments in the supplied accounting.", "decision": "A sales forecast cannot be copied directly into a fleet forecast. The adjustment terms need their own evidence or an explicit scenario assumption.", "caution": "The components of the residual are not identified. Renaming the whole residual as exports or scrappage would invent a result." }, "ev-16": { "question": "What fleet follows from a chosen future sales path and exit assumption?", "read": "Choose the assumed new-sales path and annual BEV exit rate, then inspect an end-of-year fleet share. The neighboring paths show assumption sensitivity, not likely or unlikely outcomes.", "definition": "The model fixes the total fleet at 49,486,487 cars and annual registrations at 2,857,591. It adds assumed new BEVs, removes assumed BEV exits and sets other net adjustments to zero.", "meaning": "Reaching 100% BEV new sales in 2035 produces a fleet share of about 36.2% that year under the middle exit assumption. Earlier non-BEV purchases remain part of the stock.", "decision": "Use the instrument to see which assumption moves the result and to separate a sales target from a fleet target. Replace the fixed inputs with supported alternatives before applying it to a real plan.", "caution": "Exit rates of 1%, 3% and 5% are assumptions, not estimated lifetimes. Neither the middle path nor the envelope is a probabilistic forecast." } }, "curves": { "title": "Before trusting a curve, ask what job it is doing.", "deck": "An observed line connects measurements. A fitted line compresses the past into a few parameters. A forecast extends that model beyond the observations used to fit it. The shapes can look similar, but the claims are very different.", "lineJobs": [ { "name": "Observation", "description": "A recorded share at a stated date. Connecting the points helps reading; it does not turn the gaps into new observations.", "example": "Country histories and German registrations", "href": "#ev-02" }, { "name": "Fitted function", "description": "A mathematical shape chosen to describe available observations. Its parameters are estimates or, in the teaching laboratory, explicit illustrative inputs.", "example": "Logistic and Gompertz shapes", "href": "#curve-lab" }, { "name": "Frozen forecast", "description": "A prediction made without using later target observations. Its usefulness is judged against what happened and a simple benchmark.", "example": "German December-origin forecasts", "href": "#ev-07" }, { "name": "Diagnostic profile", "description": "The fit obtained as one parameter is varied. The horizontal axis is a parameter, not time; lower loss does not mean higher probability.", "example": "The saturation-ceiling inspection", "href": "#ev-13" }, { "name": "Conditional scenario", "description": "What the stock-flow arithmetic implies if stated assumptions hold. Changing an assumption changes the result, not the historical evidence.", "example": "The 2026–2040 fleet paths", "href": "#ev-16" }, { "name": "Reader sketch", "description": "Your proposed continuation of a historical path, or a labeled starter. It is compared with the record only after an explicit reveal.", "example": "The five-year drawing exercise", "href": "#threshold-lab" } ], "models": [ { "id": "naive", "label": "Naive benchmark", "tag": "The control", "formula": "Annual: next share = last observed share", "definition": "For annual data, repeat the last share. For monthly data, repeat the last twelve-month cycle. The monthly benchmark therefore allows seasonality without estimating a new adoption trend.", "reading": "Its simplicity is the point. A more elaborate model earns its place only if its extra structure improves forecasts on the same targets." }, { "id": "logistic", "label": "Logistic", "tag": "The symmetric S", "formula": "s(t) = K / [1 + exp(−r × (t − t₀))]", "definition": "K is the long-run ceiling, r controls steepness and t₀ locates the inflection. In the unseasonalized shape, the fastest gain occurs at half the ceiling, K/2. Growth then slows symmetrically around that inflection.", "reading": "A compact description of acceleration and saturation. Its smooth, monotone path does not by itself represent a subsidy shock, product constraint or reversal." }, { "id": "gompertz", "label": "Gompertz", "tag": "The asymmetric S", "formula": "g(t) = K × exp[−exp(−r × (t − t₀))]", "definition": "K is again the ceiling and t₀ the inflection, but the fastest gain now occurs at K/e, about 36.8% of the ceiling. The approach to saturation has a longer upper tail than the logistic shape.", "reading": "The same apparent takeoff can coexist with a different slowing phase. Similar fits over a short window need not imply the same long-run timing." }, { "id": "local_trend", "label": "Local trend", "tag": "Recent momentum", "formula": "logit(s(t)) = a + b × t", "definition": "The supplied implementation fits a line to log-odds over the last five annual observations or 36 monthly observations, with monthly seasonal terms. Inverse-logit transformation returns forecasts to the share scale.", "reading": "It asks whether recent momentum is more useful than a complete S-curve. Short windows can adapt, but can also extrapolate a temporary episode." }, { "id": "break_trend", "label": "Break trend", "tag": "An estimated change in slope", "formula": "logit(s(t)) = a + b × t + c × max(0, t − τ)", "definition": "The supplied model compares an unbroken trend with candidate slope changes. The Bayesian information criterion (BIC) trades improved fit against added complexity; the knot τ is selected from the data, not assigned to a policy date.", "reading": "A fitted change can improve a description without identifying its cause. Do not rename an estimated knot a causal subsidy effect." } ], "parameters": "In the formulas, t is time, exp is the exponential function and e is approximately 2.718. For logit calculations, s is a fraction between zero and one: 20% is 0.20. The logit is log(s/(1−s)); its inverse returns to zero–one. The teaching curves instead express both K and the plotted share in percent. Errors are reported in percentage points after returning to the share scale.", "fittingBoundary": "The supplied logistic and Gompertz fits minimize squared errors on the share scale after monthly seasonal adjustment, not on logit shares. Local and break trends are fitted on log-odds. The laboratory below isolates the underlying shapes: no country coefficients, monthly seasonality or new econometric estimation enter it.", "estimation": [ { "title": "Seasonality and the fitted scale", "text": "For monthly structural curves, an intercept, time trend and two annual Fourier harmonics identify the seasonal component on the logit scale. The code removes that component, returns to shares, fits the structural curve to those adjusted shares, then adds the seasonal component back in log-odds for fitted values and predictions. Local and break trends instead include the seasonal regressors in their logit regression. Naive simply repeats the supplied prior cycle." }, { "title": "Bounds are computational assumptions", "text": "For free logistic/Gompertz fits, K lies between max(0.0001, 0.55 × the largest adjusted share) and 1, r between 0.001 and 5, and t₀ between −30 and 100 in model time. Logit calculations bound shares to [0.00001, 0.99999] for numerical stability. These bounds are estimation choices, not empirical confidence limits; the source observations are not overwritten." }, { "title": "The break is selected, not imposed by a policy story", "text": "BIC compares the unbroken trend with candidate knots. The annual series considers yearly knots with at least three observations on either side; monthly candidates are three months apart with at least twelve observations on each side. A selected knot is a statistical change in slope, not proof of the event that caused it." }, { "title": "The ceiling profile is a separate diagnostic", "text": "For each selected country/window, 60 imposed K values run from max(5%, 0.9 × the largest observed share) to 100%. K is fixed and the logistic r and t₀ are refitted at each grid point. The stored RMSE is the square root of the average squared observed-scale fitting error, expressed in percentage points. This sensitivity set is explicitly not a confidence interval." } ], "lab": { "title": "Two S-curves. The same ceiling. A different slowdown.", "question": "What changes when the shape changes, while the ceiling and fastest annual gain are held equal?", "instructions": "Choose a ceiling and a peak annual gain. The two curves share their inflection at relative year zero, but reach it at different shares. Inspect the plot with pointer, touch or arrow keys. These are illustrative shapes, not country forecasts.", "finding": "For a logistic curve, the fastest rise occurs halfway to the ceiling. For a Gompertz curve, it occurs earlier in share terms, at about 36.8% of the ceiling. Shape is an assumption with consequences.", "boundary": "Illustrative mathematical functions on a fixed zero–100% share axis. The horizontal axis is years relative to an assumed inflection, not a calendar forecast. Inputs are teaching choices, not parameter estimates or confidence limits.", "method": "To match peak growth p (percentage points per year), set logistic r = 4p/K and Gompertz r = ep/K, with K expressed as a percentage and t₀ = 0. Evaluate every quarter-year from −15 to +15; the exact-value table reports whole relative years." }, "source": { "label": "Definitions inspected in the frozen research code; teaching functions independently evaluated, not fitted", "edition": "2026-09-18", "members": [ "code/models.py", "code/analyze_models.py" ], "href": "#methods" } }, "marketComparison": { "visibleCount": 6, "defaultCountry": "Germany", "referenceOrder": [ "Norway", "China", "Denmark", "United States", "India", "Germany", "France" ], "selection": "Keep the selected country first, then take distinct reference markets in this order until six are present. The ordering illustrates contrasting levels and paths; it is not a representative sample.", "fullContext": "All 60 recorded national histories remain available through an explicit comparison-set control and the unchanged source download." }, "media": [ { "id": "powertrain", "alt": "Cutaway side view of an unbranded electric car, with a red path from its charge port to the underfloor battery, power electronics, motor and wheels.", "caption": "Conceptual BEV power path: the charge port feeds the battery; power electronics then feed the motor and driven axle. Components and routing are explanatory, not measured vehicle geometry or an engineering specification.", "source": { "path": "content/editorial/to-charge-or-not-to-charge/media/powertrain-source.svg", "sha256": "4607247e8129142d1cb768e7ac1a3b05903cbfab19e68a314d53a07d264aeb19", "rights": "Repository-authored original; no external visual source." } }, { "id": "two-clocks", "alt": "Two separate tracks: a sequence of incoming registration cards above and a large parked fleet field below, with no connecting flow widths.", "caption": "Two clocks with different denominators. Registrations are a flow of vehicles entering during a period; the active fleet is the stock present on a reference date. The separated tracks are deliberately not a Sankey, and their widths do not encode quantities.", "source": { "path": "content/editorial/to-charge-or-not-to-charge/media/two-clocks-source.svg", "sha256": "66392b4db21f305643f4b416d63d273ae109add19a50eb5685dba78447cfb8b4", "rights": "Repository-authored original; no external visual source." } }, { "id": "charging-context", "alt": "A white BMW i3 connected by cable to a public Level 2 charging station outside Automercado Guayabos in Costa Rica.", "caption": "Charging is a place-bound activity as well as a vehicle feature. This 2022 scene is in Automercado Guayabos, Granadilla, Costa Rica; it does not depict or substantiate German registration or fleet data.", "source": { "path": "content/editorial/to-charge-or-not-to-charge/media/charging-context-source.jpg", "page": "https://commons.wikimedia.org/wiki/File:BMW_i3_charging_L2_public_station_CRI_01_2022_3477.jpg", "download": "https://upload.wikimedia.org/wikipedia/commons/1/14/BMW_i3_charging_L2_public_station_CRI_01_2022_3477.jpg", "bytes": 3210688, "width": 2400, "height": 1600, "mime": "image/jpeg", "sha1": "18d3d27f2e6384ba9d80b3d5f36995b580cc6cc8", "sha256": "0b8a1c1c696a6201e9ef55d905d389a1f28fe6a0129ef6be02c2273fe653d5a0", "creator": "Mariordo (Mario Roberto Durán Ortiz)", "creatorUrl": "https://commons.wikimedia.org/wiki/User:Mariordo", "dateTaken": "2022-01-14", "location": "Automercado Guayabos, Granadilla, Costa Rica", "license": "CC BY-SA 4.0", "licenseUrl": "https://creativecommons.org/licenses/by-sa/4.0", "retrievedAt": "2026-09-20", "adaptation": "Resized and encoded as responsive WebP and AVIF derivatives; no crop or retouching. Derivatives retain CC BY-SA 4.0.", "contextBoundary": "The photograph documents charging context in Costa Rica in 2022. It does not depict or substantiate German registration or fleet data." } } ] }