Nfi Whitepaper Final PDF
Nfi Whitepaper Final PDF
Nfi Whitepaper Final PDF
1.5 The transaction costs: USD 8-16 billion value waiting to be unlocked 6
1.6 A freight exchange and a freight index: Two-pronged solution 8
Chapter 2: Making a trustworthy freight index 9
Chapter 3: Using NationalFreightIndex.co.in 11
Chapter 4: Sample use case – drawing insights from NFI 13
Chapter 5: Making of NFI – overview of computation methodology 16
5.1 Representation 16
5.2 Granularity 16
5.3 Selection of lanes 17
5.4 Computing the base level index rates 18
5.5 Computing the aggregated index rates 19
Table of figures
Figure 1 : Schematic representation of major players 5
Figure 2 : Value unlock potential in spot marketplace (USD BN, 2018) 7
Figure 3 : Design principles for a trustworthy freight index 9
Figure 4 : Value unlock potential in spot marketplace (USD BN, 2018) 11
Figure 5 : Fortnightly articles on insights from data and field 12
Figure 6 : All-India freight indices for FY19 across truck types 13
Figure 7 : Correlation of All-India indices with macroeconomic indicators (FY19) 13
Figure 8 : Profitability indices by truck type for FY19 14
In 2018, the Indian road freight market was estimated to be at USD 150-160 bn. About USD
130-140 bn of this is the full truck load market and within this, the spot freight market is about
USD 110-130 bn. The spot freight market is growing at 9-10% per year compared to the overall
GDP growth rate of 7%. The spot freight marketplace comprises four major actor groups.
This model increases uncertainty and risk for both sides. While waiting longer gives time to
negotiate better rates, it also increases the chances the load or truck will get booked by the
time a satisfactory price is found, as each spot order comes with an urgency. Fleet owners and
transporters need to engage with multiple brokers or work with sub-optimal prices. The need
to network with more brokers is even more pressing for fleet owners with less than 5 trucks
(75% of truck owners are in this category).
The traditional marketplace has been effective majorly due to lack of a better alternative. It
works, but with inefficiencies. The transaction costs in this ‘traditional’ marketplace, including
brokerage, inventory and fleet underutilization are about USD 8-16 bn per year, which is about
7-14% of the total marketplace size.
The information asymmetry of location pair and vehicle type is a deterrent to creation of a
composite index, which could reflect the performance of the spot freight market in the country.
Information is disparate with low refresh frequency, disseminated only with press releases,
limited to few key lanes, marred with lack of specificity and gathered through select sources
where the risk of a sampling bias is high.
The industry needs a two-pronged solution. First, a digital freight rate exchange which reflects
the transaction prices across millions of lanes in the country. Second, a freight index which
Trustworthy indices worldwide have clear markers. They transparently reflect the underlying
logic and the logic remains consistent over a period of time. The data on which they are based
has high credibility as they are more of actual transactions and less of imputations. The indices
are updated regularly at a stated frequency. The indices cover the market segments
comprehensively and often come with granular decomposed sub-indices to reflect micro-
segments. Each micro-segment is well represented through appropriate sampling and
weighting. There is a universal appeal to how indices are projected and interpreted across
countries and user groups, which makes them compliant and scalable for a variety of use
cases.
With these rationales and accompanying design principles, the Rivigo Rate Exchange (RRE)
was created. Further, the National Freight Index (NFI) has been launched based on the prices
reflected in the exchange. While RRE gives a live spot rate on over 7 million lanes (origin,
destination) and vehicle type combinations in the country, the NFI is a barometer of the road
freight spot market, offering an aggregated picture of both, live rates and historical trends of
spot prices across 150 different combinations.
On the page nationalfreightindex.co.in, one can look at the live freight index across adjustable
filters on truck types, origin zones and haulage distance and lane-wise live spot rates.
Additionally, a 13-month index history, a dedicated page on insights generated from indices,
and a 6-month price history for top lanes are also available. Over the next few months, new
features will be launched to enable index personalization and reflect market mood live.
NFI can cater to use cases for logistics decision makers, researchers, sales force of trucking
related businesses (OEMs, NBFCs), researchers, academicians and students. One can use the
indices in conjunction with their own data points or refer to the insights generated by the NFI
team. Below are some interesting and useful insights
- Container truck indices are more sensitive to fuel price variation vs Open truck indices.
- Using indices and fuel prices one can infer that fleet owner profitability took a hit in Q3
FY19 and is recovering now.
- Inflation indices (CPI, WPI) could be useful predictors of freight indices for the following
month. Open truck indices follow WPI while Container truck indices follow CPI.
The underlying data used for computing the indices is based on Rivigo’s marketplace
transactions and machine learning and economics principles powered algorithms that use
millions of data points. The algorithm is continuously learning and will only get better with
time and transaction volumes. The journey towards creating a truly transparent digital
marketplace has only started. It is for the industry to unlock the multi-billion-dollar
opportunity that lies ahead.
65% of the annual billion-ton-km (BTKM) moved across the country is by road. Since bulk of
the rest of 35% is rail, and given road is almost 1.8 to 2 times as expensive a mode as rail, road
freight constitutes almost 75% of the total transportation spend, which is roughly USD 150-
160 bn for 20183. Approximately USD 20 bn out of this is the time sensitive and part-truck-
load freight market, leaving USD 130-140 bn for regular full truck load (FTL) freight
market4.
• Agriculture and MSMEs form about 50% of the spot freight (USD 60-70 bn).
o Spot trucking can largely be attributed to agriculture and industry, which
constitute 38% of India’s GDP (15% and 23% contributions respectively)5. About
58% of the GVA in the industry sector is on account of manufacturing.
o 25% of the road freight spend (USD 30-35 bn) in the country is by the
agriculture sector6. Due to very short shelf lives of the products moved and the
seasonality of the industry, almost the entire agricultural produce
transportation is done on spot bookings.
o Rest of the 75% of the road freight is by the manufacturing sector. Micro, small
and medium enterprises (MSMEs) contribute 6% to the manufacturing GDP of
India by value, but as high as 34% by volume7. MSMEs, due to their low relative
production volumes and drop sizes, operate on spot markets for freight
bookings. Hence, this adds another 25% of the road freight (USD 30-35 bn)
spend to spot transactions.
• In the remaining 50% of spend (driven by large size manufacturing firms)
transportation partners typically operate with owned and market fleet to meet client
guidelines on service levels. A large sized steel maker can mandate transporters to have
up to 30-40% owned fleet. However, rarely does a transporter have both onward and
return leg contractual agreements8. Transporters for such large size firms rarely have
more than 15-20% of total ton-km booked on contractual basis. Hence, conservatively,
at least 80% of truck bookings (USD 50-60 bn) in this segment are booked on spot
(excluding specialized vehicles such as tankers and dumpers)
I. End customer / shipper: Provides loads to be moved on a day to day basis. They could
be manufacturers, traders or retailers or any actor which owns the goods to be moved.
The loads provided come along with load tonnage, preferred truck type, loading and
unloading point(s), an expected freight rate and sometimes loading instructions.
II. Transporter: Takes the operational and financial risk of moving the goods from the
loading to the unloading point. He is responsible for truck placement, paperwork,
payments to supplier, tracking of goods, collection from end customer upon
submission of proof-of-delivery. If he has entered into a contract (fixed price for agreed
lanes usually lasting 6 months to 3 years with fuel price variation clauses) with the end
customer, he might be mandated to serve a certain percentage of loads (typically 30-
40%) with his owned vehicles in order to ensure placement service levels. The rest of
the onward loads and almost all of the return loads are served through the spot
markets. Hence, the transporter needs to place his owned as well as market sourced
trucks and manage with contracted rates or run the risk of losing the share of wallet of
the client in the contract renewal process. Setting the right contract rates early
becomes imperative for the transporter to stay afloat.
III. Broker: Transporters or sometimes end customers directly reach out to these agents
to source supply. Usually a single-person shop setup in a truck mandi11, the booking
agent ties up with or ‘attaches’ himself to anywhere between 20-150 trucks spread
IV. Fleet owner / carrier: They own trucks and typically engage 1-2 drivers per vehicle.
75% of India’s trucking inventory is with those who own up to 5 trucks. Only 10% is
with fleet operators owning 20 or more trucks. Small fleet owners or single truck driver
cum owners are hugely dependent on these agents to provide loads and keep their
truck utilization up.
Most loads require that a broker confirm the truck details within 1-3 hours of request, even if
the loading is scheduled many hours away or on the next day. This leads to a tricky situation
for both demand and supply sides. While waiting longer gives time to negotiate for better
rates, it also increases the chances the load or truck will get booked by the time a satisfactory
price is found, as each spot order comes with an urgency.
1.5 The transaction costs: USD 8-16 billion value waiting to be unlocked
The industry bears the brunt of these inefficiencies in the following ways.
• Brokerage cost (A): Brokerage fees are typically 3-8% of the spot rate. About USD 3-
6 bn is spent annually on facilitating these transactions. This amount underscores the
importance brokers play in the freight marketplace, especially in getting loads to small
fleet owners who have no other trustworthy means of accessing loads. Contrast this
with equity brokerage firms who charge anywhere between 0.01 to 0.5% brokerage
fees or in real estate where 1% of deal value as brokerage is the norm.
• Inventory holding cost (B): On and above the one-day planning time already built in,
delays in getting the load or the truck, irrespective of the price, adds to the delays. On
the demand side, costs for 0.5 to 1 day of additional inventory are to the tune of USD
2-4 bn.
110-120
2.5 - 5 8 - 16
2 -4
3 -6
Brokerage cost Inventory holding cost Truck underutilization cost Total transaction cost Total market size
A B C D = A+B+C
Hence, the industry ends up paying about USD 8-16 bn as transaction costs (A, B and C).
This is at least 7-14% of the entire trucking spot market in the country.
Additionally, there are costs related to information asymmetry on pricing. In order to immune
oneself from spot rate swings, both demand and supply sides end up leaving value at the
table. The willingness to pay on the demand side is typically 2-4% on the higher side and
suppliers are often willing to work with 2-4% discount in case trucks are heading towards their
base locations on a return leg. Almost 4-8% of deal value or USD 4-10 bn is lost on account
of lack of price transparency.
The clock is ticking as the industry is growing at 8-10% per year. This problem can be
addressed through a trustworthy freight exchange which removes the information asymmetry
at scale. The exchange should reflect day to day movement in prices across majority of lanes
and truck types in the country, empowering even the smallest fleet owner in the remotest of
mandis.
The lack of price transparency not only affects the parties directly involved in spot transactions
but also policy and decision makers across government bodies, financial institutions and
manufacturing industries impacted by the performance of the freight industry. Until now, there
has been no composite index which reflects the performance of the spot market in the country,
that is available for all and at all times. Information is disparate, disseminated only with press
releases, limited to few key lanes, marred with lack of specificity and gathered through select
• CV OEMs can cut down their forecasts and focus on after sales for the next quarter
Both, the index and the exchange, are based on machine learning and economics
powered pricing algorithms which are continuously learning and improving. The rates on the
exchange and the indices are computed using millions of data points from historical
transactions, proprietary sources and crowdsourced data with the ultimate purpose of giving
a fair and honest representation of the state of the spot market in the country.
Imagine a regular user of any trustworthy index, who performs real financial transactions and
presents insights to his/her company’s management using the index – a number or a trendline
which represents the state of affairs of the indicator being measured. A lot rides on the index,
especially when organizations are going to take decisions based on the same. Taking cues
from several indices such as Brendt Crude Oil Prices, Index of Industrial Production, Baltic Dry
Index, SENSEX and others – a freight index should have the following characteristics in order
to be trusted and adopted.
II. Credibility: The index should be reflective of actual indicators (prices, rates, volumes)
and hence one should use more of actual than imputed data.
III. Regular refresh: The index needs to be updated on a periodic basis. Thus, only those
data sources and categories with regular updates should be used. For example, one
cannot use routes or markets where the transaction happens once a month for a daily
refresh. The more frequent the update – less the number of data points per update.
Hence, one should be cognizant of the trade-offs and perform refreshes accordingly.
IV. Consistency in logic: Lanes used, weights assigned, granularity should be maintained
period on period. This implies that one cannot change the lanes, weights and
granularity for a given time series of index. In the future, one can start off a new series,
if based on or started from a different reference month.
V. Usability: The index should be useful to the user in the maximum possible way with
least additional assumptions. Hence the index should be
o Granular: The granularity should be such that each user can use the most
granular index to infer the spot market movement for his own volume / network
mix.
VI. Universality: The index should be universally accepted and understood. Take the case
of Container, 32ft mxl, ex North for 750+ km haulage. It is difficult to represent trends
on the entire population of lanes with only trip prices or price per ton. Hence, the
trends of freight rates on this combination needs to be represented in INR per ton-km
as it is universally accepted and can easily be used by users for their own network
footprint.
VII. Scalability: The index should meet all bars on credibility and regulatory requirements
for any new developments, now or in the future, built upon the index. For example, if
a derivative product is built on the index, the underlying data should be reflective of
highly liquid transactions and meet any legal criteria required.
These principles have been adhered to while designing and building the National Freight
Index.
The page hosts the freight indices and many other useful features for visitors, as discussed
below.
1. Filter tray: A user can view indices at 150 different combinations of truck type, origin
zone and haulage distance depending on the use case.
2. Live section: This portion gives users the live rates at the granularity chosen through
the filters. The same is represented in the pegged form with the base month as Apr-
18. A comparison with the previous month and the same month last year is also shown.
3. History section: Here a 13-month historical is available trend and users can overlay
fuel prices relevant to the filters chosen to draw insights.
4. Top lanes: This section gives a view of the prevailing market rates and a 6-month price
history on some of the key lanes referenced to compute the index for the particular
selection of truck type, zone and distance.
5. Rivigo Rate Exchange (RRE): The ticker gives users complete access to live full truck
load rates across 7 million origin, destination, vehicle types in the country.
Under the ‘insights’ section, users can also read through the NFI team’s perspectives on the
National Freight Index, and some of the interesting observations found by analysing NFI along
Users get an exhaustive experience while browsing through the NFI website. Going forward,
several more insightful features will be launched. One such feature is where enterprises can
customize the weights used, build an index which is on the lines of their own business footprint
and can simulate any potential savings on their road freight bill by moving from contract to
spot rates. Another such feature is an on-ground mood reflector, broadcasting sentiments of
users and their responses to specific questions deemed relevant, based on patterns observed
on NFI.
The National Freight Indices’ monthly averages at a pan-India level for all truck types (open,
container & trailers)13, were compared against monthly averages of several other market
indicators.
I. Fuel
II. Inflation
III. Interest rates
IV. Industrial production
V. Stock market sectoral and composite indicators
VI. Other lead indicators e.g. Purchasing Managers’ Index
Going forward, an outlook on the coming month for both the freight and its related industries
will be shared on the RRE, based on the predictive potential of the indicators studied so far.
For example, the commercial vehicles (goods) industry within auto sector should be specifically
interested in the profitability indices – a measure of net income gathered by each segment,
proxied by normalizing freight indices with fuel and interest rate expenses. Fuel is assumed to
be 45% of the total freight bill, with EMI 30% of the interest expense. Any other factor costs
influenced by inflation (e.g. driver salaries, maintenance) will not be reflected month on month.
Hence,
Where,
• This was reflected in the 20% decline in sales of Medium and Heavy Commercial
Vehicles (MHCV) segment of top CV manufacturers in the country for Nov 2018.
• After the increase in loadable capacity after the gross vehicle weight (GVW) revisions,
truckers have crossed the profitability levels of Apr 2018 in Feb 2019. Hence, FY20 Q1
can be expected to have good y-o-y growth in the MHCV segment.
This section outlines the methodology adopted to compute the National Freight Index. The
design principles discussed earlier were adhered to make NFI a comprehensive and
trustworthy index.
5.1 Representation
The National Freight Index is represented in two major forms. Firstly, in INR per ton-km and
secondly, as a figure pegged to a base month. April 2018 is considered the base month. This
is how most indices are represented, in a purist as well as a pegged form.
The National Freight Index is computed live and is refreshed every hour on Rivigo’s Freight
platform. Along with this, average indices for a month are published for the previous 13
months. In order to further analyze index trends, avg. fuel rates for the month are available for
overlaying on historical trends. The methodology for computation of each of these are
explained in a later section.
5.2 Granularity
Fundamentally, the NFI is a composite of several sub-indices. In some places, few cohorts have
been combined to aggregate more data points for computational reasons. The granularity
available on the NFI page is as follows:
o Pan-India
o North
o South
o East
o West
• The origins and destinations of lanes must be in the list of locations with high road
freight traffic thoroughfare in the country. For the same, 120 locations were selected,
30 in each zone, which contributed to 65-70% of the traffic in the country. The traffic
information was fetched using Rivigo’s internal traffic volume information based on
trips, load posts and load requests and toll-booth transactions at 300+ booths across
the country during the period July to Dec 2018.
• The lanes must cover all 4 travel directions in each zone i.e. towards North, South, East
and West from any zone. This was done through visual inspection.
• There must be at least 10-20 lanes for each of the 56 combinations
• The lanes must have a decent trip or load post history on Rivigo’s Freight platform,
enough to give at least 3 data points in each month per lane.
As stated earlier, users of the National Freight Index can view the same at the most basic level
of truck type, zone of origin and haulage distance. Let this level be BL (stands for base level).
The objective is to compute the monthly average INR per ton-km at this level. This is done by
rolling up indices from two more granular levels. Suppose,
• For a particular vehicle type (Vt), the base level (BL) could be represented as: Container
32 ft mxl, from North, for 750+ km.
• Each BL is composed of 2-4 District Pairs or DPs: DP1, DP2, DP3, DP4, and so on.
• Each DP is composed of 4-10 Lanes: say DP1 is composed of Lane1, Lane2, Lane3, Lane4
and so on.
For month M, DP1 is computed as,
∑𝑝𝑖=1(𝐷𝑃 𝑖 ∗ 𝑤𝑒𝑖𝑔ℎ𝑡𝑎𝑔𝑒 𝑖 )
∑𝑛𝑖=1(𝑤𝑒𝑖𝑔ℎ𝑡𝑎𝑔𝑒 𝑖 )
Where p is the number of DPs within BL and weightage is the weight attributed (explained
later)
Live rates
The real time rates or indices are computed with a method similar to the one stated earlier.
Again, suppose,
• For truck type, Vt, base level is BL: e.g. Container 32ft mxl, from North, 750+ km.
• Each BL is composed of 2-4 District Pairs or DPs: DP1, DP2, DP3, DP4, and so on.
• Each DP is composed of 4-10 Lanes: say DP1 is composed of Lane1, Lane2, Lane3, Lane4
and so on.
At any given time, DP1 is computed as,
∑𝑝𝑖=1(𝐷𝑃 𝑖 ∗ 𝑤𝑒𝑖𝑔ℎ𝑡𝑎𝑔𝑒 𝑖 )
∑𝑛𝑖=1(𝑤𝑒𝑖𝑔ℎ𝑡𝑎𝑔𝑒 𝑖 )
Where p is the number of DPs within BL and weightage is the weight attributed (explained
later).
Suppose, for a truck type, Vt, the base level is BL. For example, Container 32 ft mxl, from North,
for 750+ km.
For any rolling up, to compute an aggregate index, the following logic is used:
For example, to arrive at index for Container 32 ft mxl, from North, the following logic will be
used:
[(Index value Container 32 ft mxl, from North, for 750+ km * weightage Container 32 ft mxl, from North, for 750+ km)
+
(Index value Container 32 ft mxl, from North, for 250-750 km * weightage Container 32 ft mxl, from North, for 250-750 km)]
/
(weightage Container 32 ft mxl, from North, for 750+ km + weightage Container 32 ft mxl, from North, for 250-750 km)
These weightages are planned to be kept unchanged with time and if changed will be
communicated to all concerned users. NFI team can be contacted to get more information on
the weightages.
INR per tonkm for the current time period under considered
𝐼𝑛𝑑𝑒𝑥 (𝑝𝑒𝑔𝑔𝑒𝑑) =
INR per tonkm for the month of April 2018
Acknowledgements
The NFI team would like to thank and acknowledge the following Rivigo owners who have
contributed in various phases of NFI development.
Siddhant Mishra
Shreynik Kumar
Vipin Chander
Narendra Sisodiya
Bharathi Rajan Muthu Krishnan
Yogesh S
Akshay Agarwal
Romit Choudhary
Rishabh Tiwari
Jaiprakash
Nikhil Ranjan
Tushar Makkar
Sourabh Kalal
Satish Kumar
Vipul Sharma
Gaurav Sharma
Shobhit Sood
1
The Hindu Businessline, “India’s GDP to grow at 6.9 per cent in FY19, says Ind-Ra”, 27 May 2019
2
Livemint, “Debunking-Indias-logistics-myths”, 23 Mar 2018
3
NTDPC, Planning Commission, “India Transport Report, Moving India to 2032”, 2014
4
Rivigo analysis
5
Ministry of Statistics and Programme Implementation, 2018-19
6
McKinsey report in Livemint, “Debunking-Indias-logistics-myths”, 23 Mar 2018
7
Ministry of Micro, Small and Medium Enterprises, Govt. of India, Annual Report 2017-18
8
Rivigo analysis
9
Economic Survey, Govt. of India, 2017-18
10
Central Statistics Office, India, First Advance Estimates of GDP 2018-19
11
Truck Mandi is an offline micro-market of trucks on spot, about 230 of them exist in the country
12
Rivigo analysis, voice of customer
13
National Freight Index