Advanced 16 min read Module 23

How Risk and Route Indices Are Built

Full technical documentation of the Route Disruption Index (RDI), Household Freight Risk Index (HFRI), and Consumer Shipping Sensitivity (CSS). Data sources, weights, computation methods, z-score methodology, and component breakdowns.

Why We Build Our Own Indices

Existing freight indices -- the Baltic Dry Index, the Freightos Baltic Index, the Shanghai Containerized Freight Index -- each measure a specific thing well. The BDI tracks dry bulk rates. The FBX tracks container rates. The SCFI captures Chinese export lane pricing. But none of them answers the question that matters to a household, a portfolio manager, or a municipal budget planner: what does this mean for the prices I pay?

Risk and Route's three proprietary indices are designed to bridge the gap between raw freight market data and its economic consequences. The Route Disruption Index (RDI) measures the severity of active shipping disruptions in real time. The Household Freight Risk Index (HFRI) estimates the probable consumer price impact of current freight conditions over the next 30 to 120 days. The Consumer Shipping Sensitivity (CSS) score classifies product categories by their vulnerability to shipping cost changes. Together, they translate the shipping market's language into terms that non-specialists can act on.

All three indices are constructed from publicly available data sources. We use no proprietary trading data, no paid broker feeds, and no insider information from shipping companies. Everything that goes into our indices can be independently verified by anyone with a FRED account, a MarineTraffic subscription, and a Baltic Exchange data terminal. This is a deliberate design choice. An index that depends on private data cannot be audited, and an index that cannot be audited has no claim to authority.

Route Disruption Index (RDI): Technical Specification

The RDI is a composite score from 0 to 100 that measures the current severity of shipping route disruptions worldwide. A reading of 0 means no detectable disruption on any monitored route. A reading of 100 would indicate simultaneous severe disruptions across all major chokepoints with extreme rate acceleration and market stress -- a condition that has never occurred in the data we track but would correspond roughly to a global shipping blockade. In practice, the RDI has ranged from 4 (stable baseline, late 2019) to 78 (peak of combined Red Sea and Panama Canal disruptions, January 2024).

The index is computed daily and published on our Indices page. It is built from six weighted components, each expressed as a z-score before aggregation.

Z-Score Methodology

Each RDI component is converted to a z-score: (observed value - rolling mean) / rolling standard deviation. This standardization allows components measured in different units (vessel counts, dollars per tonne, percentage deviations) to be combined on a common scale. Rolling windows vary by component (30 to 180 days) based on the natural variability of each data source. The weighted sum of component z-scores is then mapped to the 0-100 scale using a logistic transformation calibrated against historical disruption events.

Chokepoint Transit Volume

25%
Source: AIS vessel tracking (MarineTraffic, VesselFinder)
Frequency: Daily

Measures daily vessel transits through the five critical chokepoints: Suez Canal, Strait of Malacca, Panama Canal, Strait of Hormuz, and the Turkish Straits. Decline in transit count relative to 90-day rolling average signals active disruption. Weighted by vessel type (container > tanker > bulk).

Baseline = 90-day rolling mean. Each daily reading is converted to a z-score. Values below -1.5 trigger an 'elevated' flag; below -2.5 triggers 'severe'.

Route Deviation Index

20%
Source: AIS-derived voyage data, shipping line schedule data
Frequency: Daily

Tracks the percentage of vessels on affected trade lanes deviating from their standard route. During the Red Sea crisis, this component captured the shift from Suez to Cape of Good Hope before rate indices reflected the cost impact. Measures actual sailing distance versus optimal-route distance for matched vessel pairs.

Baseline = 180-day rolling mean deviation percentage. Z-score calculated against historical distribution. Values above +2.0 indicate significant rerouting.

Freight Rate Acceleration

20%
Source: Baltic Exchange (BDI sub-indices), Freightos Baltic Index (FBX), Drewry WCI
Frequency: Daily (BDI), Weekly (FBX, WCI)

Captures the rate of change in freight rates, not the level. A market at $3,000/FEU that was at $1,500 a month ago scores higher than a market stable at $5,000/FEU. This distinguishes disruption-driven spikes from structurally high rate environments.

30-day rate of change expressed as a z-score against 2-year historical distribution of 30-day changes. Separate z-scores for container (FBX), dry bulk (BDI), and tanker (BDTI) -- then weighted 50/30/20.

FFA Curve Stress

15%
Source: Baltic Exchange FFA settlements, CME container futures, BDRY ETF holdings
Frequency: Daily

Measures the degree of backwardation or contango in the freight forward curve. Steep backwardation during a disruption indicates the market expects the event to resolve; contango during a disruption indicates expectations of sustained tightness. The signal is the deviation from normal curve shape, not the absolute level.

Spread between 1-month and 6-month FFA expressed as z-score against 3-year historical spread distribution. Separate calculations for Capesize, Panamax, and container routes.

Insurance and War Risk Premium

10%
Source: Lloyd's Market Association Joint War Committee listed areas, broker-reported war risk premiums
Frequency: Weekly (JWC list changes are event-driven)

War risk insurance premiums spike when underwriters assess elevated threat in a shipping zone. The JWC listed areas designation is a binary trigger (listed or not), while the premium percentage is continuous. Both feed into this component.

War risk premium as percentage of hull value, converted to z-score against 5-year distribution. JWC listing changes are treated as step-function events adding 0.5 to the raw z-score.

Port Congestion

10%
Source: AIS-derived vessel dwell time at anchorage, port authority throughput reports
Frequency: Daily (AIS), Monthly (port reports)

Counts vessels waiting at anchor outside major ports and measures average dwell time. Congestion at destination ports extends effective voyage time, reducing available vessel supply even without a chokepoint disruption. This component captured the 2021 Los Angeles/Long Beach backup months before it peaked.

Vessel-days at anchorage above normal berth availability, expressed as z-score against 2-year rolling baseline for each port cluster. Aggregated across 20 major port complexes.

Household Freight Risk Index (HFRI): Technical Specification

The HFRI answers a different question from the RDI. Where the RDI measures the disruption itself, the HFRI estimates the downstream economic consequence: how much are current freight market conditions likely to affect consumer prices over the next one to four months? The HFRI is expressed as a percentage above or below baseline household freight cost exposure. An HFRI of +12% means current conditions are estimated to add 12% to the freight-related component of household spending relative to the 5-year average.

The HFRI is computed weekly and updated on Mondays. It draws on both shipping market data and macroeconomic indicators.

Container Rate Pass-Through Estimate

30%
Source: FBX, SCFI, BLS Import Price Index (FRED series MIPNSA)
Frequency: Weekly (rates), Monthly (import prices)

Estimates the proportion of current container shipping costs that will flow through to retail consumer goods prices. Uses a 90-day lag model calibrated against 2018-2025 data, where container rate changes explained 12-18% of core goods CPI variance with a median lag of 97 days.

Calculation: Current FBX level relative to 5-year average, multiplied by historical pass-through coefficient (0.14), adjusted for inventory coverage ratio.

Energy Freight Cost

25%
Source: Baltic Dirty Tanker Index (BDTI), Henry Hub natural gas (FRED), WTI crude (FRED), EIA import data
Frequency: Daily (BDTI, commodity prices), Monthly (EIA)

Captures the freight cost component of energy delivered to US consumers. Tanker rates affect the landed cost of imported crude oil, which feeds through to gasoline and heating fuel. LNG shipping costs similarly affect natural gas prices in import-dependent markets.

Calculation: BDTI z-score weighted by US crude import share (currently ~35% of consumption), plus LNG spot freight rate deviation from 2-year mean.

Agricultural Commodity Freight

20%
Source: BDI Panamax and Supramax sub-indices, USDA export reports, FRED agricultural commodity series
Frequency: Daily (BDI), Weekly (USDA), Monthly (FRED)

Grain, fertilizer, and oilseed shipping costs as a proportion of delivered commodity prices. When Panamax rates spike on US Gulf-to-Asia grain routes, the freight cost as a percentage of the FOB grain price increases, and this eventually passes through to food CPI with a 60-90 day lag.

Calculation: Panamax/Supramax rate z-score multiplied by freight-as-percentage-of-commodity-value for wheat, corn, soybeans, and fertilizer (weighted by US import/export share).

Disruption Duration Multiplier

15%
Source: RDI (internal), historical disruption duration database
Frequency: Daily

Adjusts the household impact estimate based on how long a disruption has persisted. A one-week rate spike has minimal consumer impact because importers hold inventory buffers. A three-month sustained disruption triggers contract renegotiations and repricing. This multiplier scales from 0.3 (first week) to 1.0 (beyond 90 days).

Calculation: Piecewise linear function: 0.3 at day 1, rising to 0.6 at day 30, 0.85 at day 60, and 1.0 at day 90. Remains at 1.0 for disruptions beyond 90 days.

Inventory Buffer Offset

10%
Source: Census Bureau wholesale inventories (FRED series WHLSLRINSA), ISM PMI inventories component
Frequency: Monthly (Census), Monthly (ISM)

High inventory levels delay the transmission of shipping cost increases to consumers because retailers and wholesalers can draw down existing stock purchased at lower freight costs. This component acts as a dampener: when wholesale inventory-to-sales ratios are elevated, the HFRI is reduced.

Calculation: Inventory-to-sales ratio z-score (inverted: high inventories reduce the index). Capped at -0.5 z-score impact to prevent excessive dampening.

Consumer Shipping Sensitivity (CSS): Category Scores

The CSS is not a time-varying index like the RDI or HFRI. It is a static classification that assigns each major consumer product category a sensitivity score from 0 to 1, indicating how strongly that category's retail price responds to changes in shipping costs. A score of 1.0 would mean perfect pass-through -- every dollar of shipping cost increase appears in the retail price. A score of 0 would mean complete absorption -- shipping costs change but retail prices do not.

CSS scores are estimated from regression analysis of shipping cost indices against CPI sub-category indices over the 2015-2025 period, controlling for exchange rates, commodity input prices, and domestic labor costs. They are updated annually as new data becomes available.

Category Sensitivity Lag
Gasoline and diesel High (0.82) 7-14 days
Fresh produce (imported) High (0.71) 14-30 days
Electronics and appliances Moderate (0.38) 90-120 days
Clothing and footwear Moderate (0.33) 120-180 days
Automobiles and parts Low-Moderate (0.28) 90-150 days
Furniture and household goods Moderate (0.45) 60-120 days
Pharmaceutical ingredients Low (0.15) 30-90 days

The CSS scores are used within the HFRI calculation to weight the freight cost pass-through estimate by product category. They also appear in our Shelf Signal alerts, where we identify which specific product categories are most likely to experience price increases from a given shipping disruption.

Data Pipeline and Computation Schedule

Index computation follows a fixed schedule designed to balance timeliness with data reliability.

1
Daily, 06:00 UTC: AIS data ingestion. Vessel transit counts for all five chokepoints updated. Port congestion vessel counts refreshed. Route deviation calculations run for all tracked trade lanes.
2
Daily, 14:00 UTC (after Baltic Exchange publication): BDI sub-indices, BDTI, BCTI, and FFA settlement values ingested. Freight rate acceleration z-scores recalculated. FFA curve stress component updated.
3
Daily, 15:00 UTC: RDI composite score computed. All six component z-scores aggregated with fixed weights, mapped to 0-100 scale, published to website. Historical RDI time series updated.
4
Weekly, Monday 16:00 UTC: HFRI computed. Weekly FBX data, latest monthly BLS import price and PPI data, Census inventory data merged. All five HFRI components calculated and aggregated. Published to website with weekly change and 4-week trend.
5
Annually, January: CSS scores re-estimated using updated regression coefficients with the prior year's data incorporated. Category definitions reviewed against BLS CPI item structure changes.

Known Limitations and Model Uncertainty

These indices are estimates, not measurements. The distinction matters. The BDI measures actual freight rate assessments collected from working brokers. The CPI measures actual prices collected from retail establishments. Our indices estimate the relationship between maritime freight conditions and consumer price outcomes. Every estimate carries uncertainty, and we are explicit about where that uncertainty is largest.

Sources of Model Uncertainty

  • Pass-through coefficients are unstable. The relationship between freight rates and consumer prices varies by economic cycle, exchange rate regime, and inventory conditions. The 0.14 container-to-CPI coefficient estimated from 2018-2025 data may not hold in the next cycle.
  • AIS data has coverage gaps. Satellite AIS coverage is excellent in open ocean but can miss vessels in congested coastal zones. Terrestrial AIS is dense in developed-nation waters but sparse in parts of Africa, South America, and Southeast Asia. Our chokepoint transit counts may undercount by 3-8% depending on location.
  • Inventory data lags. The Census Bureau wholesale inventory report is published 6-8 weeks after the reference month. Our inventory buffer offset in the HFRI uses the most recent available data, which may not reflect current conditions during rapid destocking or restocking episodes.
  • Novel disruption types. Our z-score baselines are calibrated against historical data. A disruption type not seen in the calibration window (cyberattack on a major port operating system, for example) could produce misleading z-scores because the historical distribution does not include that scenario.

We publish confidence intervals alongside point estimates where the data supports it. The HFRI, in particular, is reported with a range (e.g., +8% to +15%) rather than a single number when component z-scores diverge significantly, indicating that the underlying signals are giving conflicting readings about the likely consumer impact.

Key Takeaways

  1. The RDI (0-100 scale) measures active shipping disruption severity from six components: chokepoint transit volume, route deviation, freight rate acceleration, FFA curve stress, insurance premiums, and port congestion. It is computed and published daily.
  2. The HFRI estimates the consumer price impact of current freight conditions over 30-120 days, expressed as a percentage above or below baseline household freight cost exposure. It is computed and published weekly.
  3. The CSS assigns static sensitivity scores (0 to 1) to consumer product categories, indicating how strongly each responds to shipping cost changes. Gasoline and fresh produce score highest; pharmaceuticals score lowest.
  4. All three indices use z-score standardization to combine data measured in different units on a common scale. Rolling baselines prevent seasonal patterns from distorting the signal.
  5. Every data source is publicly available and independently verifiable. No proprietary trading data or private broker feeds are used in any index calculation.
  6. The HFRI's container-to-CPI pass-through coefficient (0.14) and the CSS regression scores are empirical estimates that carry meaningful uncertainty. We report ranges where signals conflict.
  7. Index computation follows a fixed daily and weekly schedule tied to Baltic Exchange publication times and BLS data release dates.

Further Reading

This module was produced by Risk and Route with AI assistance and human editorial review. Index methodologies are subject to periodic revision as new data sources become available and calibration windows are updated. Nothing on this site constitutes investment advice. Index readings are analytical estimates, not measurements. Consult a licensed financial adviser before making investment decisions.

Freight Market Intelligence, Weekly

Free weekly analysis including RDI and HFRI readings, component breakdowns, and what the indices signal for consumer prices.

Subscribe Free