What we measured. Country-level new registrations (demand) from ACEA and vehicle production (supply) from OICA. Each country sheet closes apparent consumption and rolls up to regions; regions are softly aligned to OEM geography disclosures without violating price/capacity corridors.
Core identity (country closure):
C_c,t ≈ Q_c,t + M_c,t − X_c,t
Value math (per country):
V_c,t = U_c,t × P_c,t (Units × ASP)
Regional soft‑anchor (constrained least squares) keeps country logic intact while nudging to disclosure bands.
Checks we ran:
| Sources | ACEA registrations (country tables), OICA production by country; national customs portals for M/X where needed |
|---|---|
| Pull dates | Monthly/Annual (ACEA), Annual (OICA); customs as per country |
| Transforms | FOB/CIF harmonization; re-export reassignment; FX: history current USD, forecasts constant FX |
| Confidence | Tier A where closure gap ≤2%; Tier B if mirror-stat gap >2% but <5% |
| Caveat | Residual gap driven by stock change visibility in a subset of countries |
Problem. Short or volatile official series; revisions/late postings. We anchor at the last audited point and forecast missing months with Eurostat Production in Construction Index (PCI) and construction price/cost indices, then re-test value against ASP and capacity corridors.
Driver growth blend:
V̂_t = V_{t−1} × ∏_j (1 + w_j · g_{j,t}), with Σ w_j = 1, w_j ≥ 0
QC loop (client-visible):
| Sources | Eurostat PCI (volume), Construction Producer Price/Cost indices; national statistics for corroboration |
|---|---|
| Pull dates | Monthly (PCI, prices/costs) |
| Transforms | Seasonal adjustment alignment; outlier winsorization; driver weighting caps |
| Confidence | Tier B (driver-filled months) → Tier A after official revisions land |
| Caveat | Short‑term volatility handled via smoothing; revised on official restatements |
Reality. PRODCOM cells may be confidential or revised later, creating holes. We detect flags, decide fill method, and reconcile back to trade and capacity.
Flag-aware gap fill:
If flags show 'missing' with adjacent years available:
CAGR = (V_{t+1}/V_{t−1})^{1/2} − 1; V̂_t = V_{t−1} · (1 + CAGR)
If series is choppy or re-based: use driver blend (as in Case 2).
We publish the flag ('C' confidential, 'L' missing) beside the filled cell and keep confidence at Tier B until official release arrives.
| Sources | Eurostat PRODCOM tables; methodology/user manual for flags and confidentiality; UN Comtrade/ITC for trade corroboration |
|---|---|
| Pull dates | Annual (PRODCOM), Annual/Monthly (trade) |
| Transforms | SDMX flag handling; CAGR or driver blend; reconciliation to C = Q + M − X ± ΔS |
| Confidence | Tier B for filled cells; upgrade when official data arrive |
| Caveat | Where reclassification occurs, previous years are restated with an audit note |
Use case. Country bottom-up is noisy in small markets. We pull a bellwether's (e.g., TSMC) regional revenue shares as a soft anchor and nudge regional totals without violating country-level corridors.
Constraint sketch:
min Σ_r ((V_r − V̂_r^{filings})/σ_r)^2 s.t. Σ_{c∈r} V_c = V_r, V_c ≥ 0
Client sees a 'Regional alignment' chip: On‑band / Near‑band, with doc reference and pull date.
| Sources | Bellwether annual/20‑F reports for geography mix; company decks for updates |
|---|---|
| Pull dates | Annual (AR/20‑F), Interim (quarterly decks) |
| Transforms | Constrained least squares; corridor caps by country preserved |
| Confidence | Tier A if on‑band; Tier B if near‑band with documented small residual |
| Caveat | If multiple bellwethers conflict, a weighted band is used and disclosed |
Why. Brand shares and installed base drift without a clean device universe. We join vendor/product masters to FDA GUDID (via AccessGUDID), dedupe by DI, regroup families, and link recalls to survival curves (retirements).
Installed base and recall linkage:
IB_t = IB_{t−1} + Ship_t − Retire_t, with Retire_t from survival S(a) and recall shocks applied to affected cohorts
Client impact: cleaner shipment roll-ups; recall quarters show a transparent step in retirements; lineage lists DI coverage and pull dates.
| Sources | FDA GUDID (device identifiers), AccessGUDID (queries/dumps); MAUDE/adverse events where relevant |
|---|---|
| Pull dates | Quarterly (DI sync), Monthly (recalls) |
| Transforms | DI joins/dedup; family grouping; recall-to-retirement mapping |
| Confidence | Tier A when DI coverage is high; Tier B if partial |
| Caveat | Where DI mapping is partial, affected shares show wider bands |
Observation. EU apparent consumption (value) rose with flat tonnage. Diagnosis: anti‑dumping measures lifted effective import prices and altered supplier mix.
Tonnage closure holds:
C = Q + M − X (tons), while Value = Tons × ASP with ASP corridor mid adjusted for duty‑impacted origins
We mark the duty start as a vertical step on the series and raise the corridor mid for affected import lanes; brand/channel ladders are re-tested for margin consistency.
| Sources | EU trade‑defence case history for CN 39076100; Eurostat trade and PRODCOM context |
|---|---|
| Pull dates | As per case timeline; monthly trade where available |
| Transforms | Duty-effective price adjustment; corridor shift; supplier mix reweighting |
| Confidence | Tier A if duty timing and price effect are well evidenced |
| Caveat | Short‑term substitution between origins can temporarily widen bands |
FMI iDEA — Assumptions visible. No sleight of hand.
Methodology
Thank you!
You will receive an email from our Business Development Manager. Please be sure to check your SPAM/JUNK folder too.
Chat With
MaRIA