PreprintPDF Available

The end of 'set it and forget it' pricing? Opportunities for market-based freight contracts

Preprints and early-stage research may not have been peer reviewed yet.

Abstract and Figures

In the for-hire truckload market, firms often experience unexpected transportation cost increases due to contracted transportation service provider (carrier) load rejections. The dominant procurement strategy results in long-term, fixed-price contracts that become obsolete as transportation providers' networks change and freight markets fluctuate between times of over and under supply. We build behavioral models of the contracted carrier's load acceptance decision under two distinct freight market conditions based on empirical load transaction data. With the results, we quantify carriers' likelihood of sticking to the contract as their best known alternative priced load options increase and become more attractive; in other words, carriers' contract price stickiness. Finally, we explore carriers' contract price stickiness for different lane, freight, and carrier segments and offer insights for shippers to identify where they can expect to see substantial improvement in contracted carrier load acceptance as they consider alternative, market-based pricing strategies.
Content may be subject to copyright.
The end of “set it and forget it” pricing?
Opportunities for market-based freight contracts
Angela Acocella, Chris Caplice, Yossi Sheffi
Center for Transportation & Logistics, Massachusetts Institute of Technology,,,
In the for-hire truckload market, firms often experience unexpected transportation cost increases due to
contracted transportation service provider (carrier) load rejections. The dominant procurement strategy
results in long-term, fixed-price contracts that become obsolete as transportation providers’ networks change
and freight markets fluctuate between times of over and under supply. We build behavioral models of the
contracted carrier’s load acceptance decision under two distinct freight market conditions based on empirical
load transaction data. With the results, we quantify carriers’ likelihood of sticking to the contract as their best
known alternative priced load options increase and become more attractive; in other words, carriers’ contract
price stickiness. Finally, we explore carriers’ contract price stickiness for different lane, freight, and carrier
segments and offer insights for shippers to identify where they can expect to see substantial improvement in
contracted carrier load acceptance as they consider alternative, market-based pricing strategies.
Key words : Truckload transportation, freight procurement, supply contracts, indexed pricing
1 Introduction
One of the unique aspects of the truckload (TL) freight industry is the non-binding nature of the
contracts. Shippers (firms with goods that must be moved) contract with transportation service
providers, or carriers (i.e., trucking companies) to haul their freight.
Contract prices are set at the time of the shipper’s strategic transportation procurement event.
The shipper communicates its forecasted demand for each lane (pickup origin to drop-off destination
pair) with carriers upfront. However, volume and capacity commitments for both parties are not
strict contractual obligations. First, the shipper is not required to offer the expected volume to
the contracted carrier. In fact, the shipper may offer more than, less than, or even none of the
awarded volume, with no direct financial penalty. This is because forecasting the precise time and
lane of future demand for trucking capacity is not realistic, especially when projections are done
often months or even a year in advance. Moreover, forecasting errors are amplified when shippers’
end customer demand changes over the course of the transportation contract.
Second, contracted carriers are not obligated to accept 100% of the volume that is offered or
tendered to them. Responding to many shippers’ uncertain demand makes committing a truck to
arXiv:2202.02367v1 [econ.GN] 4 Feb 2022
Acocella, Caplice, and Sheffi: Opportunities for market-based freight contracts
2Article submitted to ; Please do not circulate!
a specific future time and location almost impossible, especially when commitments are expected
to be in effect for a year or more in most cases. As a result, a carrier can enter a contract with
a shipper for an expected amount of freight on a lane but uncertainties introduced from other
customers, general market dynamics, or changes in the carrier’s own business require the flexibility
to reject loads in real-time capacity allocation decisions.
It is useful to think about the interactions between shippers and their for-hire motor carriers as
two stages of TL transportation: 1) the strategic procurement process with its long-term carrier-
lane matching decisions and 2) the shipper’s operational, real-time load-carrier matching decisions
with the corresponding carrier acceptance or rejection decisions. We describe these processes further
in the following subsections.
1.1 Strategic Transportation Procurement
During the strategic procurement process (see Caplice (2007) for a detailed description), shippers
conduct a reverse auction in which they send a request for proposals (RFPs) to a group of carriers.
The RFP documents the lanes (origin-destination pairs) on which the shipper requires priced
transportation services. It also includes the expected volumes on those lanes and other service level
expectations. Carriers respond to the RFPs by bidding the price they are willing to accept to serve
the expected demand on each lane in which they are interested. The prices a carrier bids depend
on the general attractiveness of serving the origin and destination regions, attributes of the freight
itself, the carrier’s previous experience working with that shipper, how well that lane fits within
the carrier’s existing networks, and the general market environment at the time.
The shipper awards the business to one or more carriers on each lane, which are referred to as
the contracted, awarded, or primary carriers. Typically a carrier is selected because it is one of the
lowest bidding carriers on the lane and has high expected level of service (e.g., load acceptance
rates, on-time pickup and delivery, technological sophistication for communication, information
and data exchange, and payments, etc.).
Due to the possibility that primary carriers may reject loads, shippers construct a routing guide
for each lane. The routing guide is a sequential list beginning with the primary carriers followed by
a set of backup carriers. These backup carriers typically have unsuccessfully bid on the lane, thus
expressing ability and willingness to serve at least some of the demand on that lane. Most often,
these carriers had bid higher prices than the winning carrier(s). The backup carriers are entered
into the routing guide at their bid price, however, there is no contract in place between the shipper
and backup carriers. This means the backup carriers’ routing guide price is not binding. Further,
there is a much lower implicit assumption that backup carriers will produce capacity.
Acocella, Caplice, and Sheffi: Opportunities for market-based freight contracts
Article submitted to ; Please do not circulate! 3
1.2 Operational Load Tendering, Acceptance & Rejection
At the time a load needs to be moved, the shipper’s transportation management system (TMS)
tenders the load to the first carrier in the routing guide (the primary, awarded carrier) at the
contracted price. If the carrier accepts the load, it agrees to move it at the contracted price. If the
primary carrier rejects the load, the load is offered to the first backup carrier at the price denoted
in the routing guide. Not only is this price between shipper and backup carrier non-binding, it is
likely higher than the primary carrier’s price. Depending on prevailing market conditions, shippers
price escalations with backup carriers reach more than 18% above the primary carrier’s contracted
price (Acocella et al. 2020).
Like primary carriers, backup carriers may accept or reject load offers. The shipper proceeds
down the routing guide until a backup carrier accepts the load or some price or time threshold is
hit. At that point, the shipper may turn to the spot market. In this case, the shipper consults a
load board where it can post information regarding the load’s pickup, drop-off, timing, and other
requirements and view carriers who have posted currently available capacity and associated prices
on the lane.
Shippers find a carrier on the spot market for independent transactions, but there is no single
spot market price at any given time, even for spot loads on the same lane. Instead, when we refer
to the “spot market price” it is the average of a range of prices. Figure 1 illustrates this point.
Figure 1: Daily Spot Prices for Single Lane It depicts actual loads fulfilled on
the spot market from 6 shippers on
a single lane in October 2017. Each
day, there may be one, many, or
no loads that require spot capac-
ity. Each load is fulfilled at a dif-
ferent price set by the individual
carrier and depends on the carrier’s
network structure at that moment.
In the remainder of this study, we
discuss the dynamic average lane-
specific spot price, depicted in the
figure, which represents an underly-
ing distribution of individual realized spot load prices.
Acocella, Caplice, and Sheffi: Opportunities for market-based freight contracts
4Article submitted to ; Please do not circulate!
1.2.1 Contractual relationships and the best known alternative
Spot market prices are highly volatile and represent the dynamic nature of immediate balance
between trucking supply and demand. For shippers and carriers alike, the spot market represents
alternative price options to a contracted load. When trending spot prices are high relative to
contract prices, carriers may be tempted away from adhering to their contractual agreements. A
primary carrier can either uphold its contractual commitments with the shipper by accepting loads
and maintaining the relationship, or it can act opportunistically by rejecting loads and offering its
capacity on the spot market with higher expected profit. However, by reneging on its contract and
rejecting loads, the carrier risks potential future business from that shipper.
About 72% of for-hire TL freight is accepted by the primary carrier, while about 5% is fulfilled by
spot capacity (Caplice 2007). The remaining 23% is moved by backup carriers within the shipper’s
routing guide. When spot market prices are higher than average contract prices - a condition
referred to as a tight market - shippers typically pay a price premium of 35% above the contract
price if they end up using spot carriers (Aemireddy and Yuan 2019).
Even when backup carrier prices or spot market prices are below a shipper’s contracted carrier’s
price, there are substantial benefits to working with primary carriers rather than backup or spot
carriers. The shipper’s RFP is a vetting process. Over the course of weeks, sometimes months, the
shipper and its bidding carriers not only communicate expected demand, service expectations, and
pricing, but the carriers have also have demonstrated fast and simple communication processes,
and technological sophistication to integrate with the shipper’s electronic data interchange (EDI)
and payment systems.
An account manager is assigned by each of the contracting parties. This person continuously
evaluates performance over the course of the contract and is the point of contact when performance
issues arise. As a result, the value for the shipper of working with a primary carrier includes the
ease of business processes and relationship developed. Due to its transactional nature, none of these
process or communication channels are established for spot loads. Furthermore, the spot market
price is not known to the shipper in advance. Thus, both service level and actual price to be paid
to spot market carriers are largely uncertain. Therefore, it is typically in shippers’ best interest to
encourage primary carriers to accept loads, even if the spot market price is expected to be better
(i.e., lower) than contract prices.
As freight markets fluctuate between periods of over and under supply - soft and tight mar-
kets, respectively - shippers’ fixed contract prices can become stale. As a result, the amount of
freight that is accepted and moved at the original contracted price declines over the course of the
contract. To mitigate this issue and ensure contract prices remain market competitive, there has
Acocella, Caplice, and Sheffi: Opportunities for market-based freight contracts
Article submitted to ; Please do not circulate! 5
been growing interest from both shippers and carriers to explore market-based pricing into their
portfolio of TL freight contracts. Understanding carriers’ willingness to stick to (or defect from)
their contracted load price helps practitioners identify the most promising network, lane, freight,
and carrier segments for this market-driven approach.
The remainder of this paper is structured as follows. Section 2 discusses practitioners’ considera-
tions regarding index-based contracts, Section 3 offers a review of the relevant literature, Section 4
summarizes our research question and the hypotheses tested, and Section 5 describes our empirical
dataset and model specifications. We present the results of the models in Section 6 and their impli-
cations, particularly for market-based pricing consideration, in Section 7. We conclude in Section
8 with a discussion of limitations of this research and areas for further exploration.
2 Market-based Freight Contracts
For every load a contracted carrier is tendered on a contracted lane, it can either accept the load
or reject it and fill available capacity with a load on the spot market. Carriers may also have other
contracted shippers on the same or similar lanes and thus other contracted business available as
an alternative option. However, because the shipper has no knowledge of its contracted carriers’
other customers and because spot prices and contract prices move up and down together, albeit
with some lag (see Pickett (2018)), we use the average lane-specific spot market price to represent
carriers’ alternative options.
As the average price of spot market loads available to the carrier increases relative to the con-
tracted price, an opportunistic carrier with low contract price stickiness will be incentivized away
from accepting the contracted shippers’ loads. For each load offered to a contracted carrier, we
calculate its Spot Rate Differential (SRD), or how much the current lane-specific spot price is
above or below the load’s contract price, as a percentage:
SRDk ,i,j,t =(SpotP r icei,j,t ContractP r icek,i,j,t)
C ontractP ricek,i,j,t
×100 (1)
where kis the load tendered to the carrier on lane with origin iand destination jat time t.
This Spot Rate Differential is the percent difference between the contract price and spot market
price at a given time and the key metric for carrier contract price stickiness. Building off of the
literature described in Subsection 3.3 we define contract price stickiness as the rate at which the
contract price must change (relative to the current lane-specific spot price) as factors exogenous
to the contract change.
A market-based pricing strategy aims to appeal to carriers with low contract price stickiness.
Market- or index-based pricing is commonly used in industrial, agricultural, and energy commodi-
ties markets. Similar to the truckload market, these products see cyclical supply and demand
Acocella, Caplice, and Sheffi: Opportunities for market-based freight contracts
6Article submitted to ; Please do not circulate!
fluctuations. Market-based pricing helps simplify price negotiations and increase transparency for
long-term contracts between sellers and buyers, particularly when price volatility over time is a
concern (Singh et al. 2016). Interest in indexed pricing has grown in the truckload freight market
as both shippers and carriers seek ways to mitigate the risks incurred by freight market cycles.
Figure 2 demonstrates the dynamic nature of the truckload freight market. We present
the Truckload Linehaul Index reported by Cass Information Systems, Inc. - a leading indus-
try provider of information and payment solutions - and national average contract and
spot market prices from our empirical dataset from September 2015 to January 2020.1.
Figure 2: Industry Index, Spot, and Contract prices
The variations in market index
and spot and contract prices over
this time period reflect both soft
and tight market periods. Until July
of 2017, the industry experienced
a soft market, where we see low
prices and index values. Follow-
ing this time, the industry experi-
enced a very tight market, with high
prices and severely deteriorated pri-
mary carrier acceptance rates. In
fact, in the first soft market period,
average primary carrier acceptance
rates were 81.9%. This number dropped to 68.5% in the subsequent tight market period. Acocella
et al. (2020) provide quantitative justification of these market periods and, along with Aemireddy
and Yuan (2019) and Sokoloff and Zhang (2020), analyze market-driven primary carrier acceptance
As the freight market cycles between soft and tight periods, index-based pricing strategies aim
to allow contract prices to adjust automatically with changes in the broad market. Otherwise,
shippers manually adjust prices through full RFPs or “mini-bids” (targeted carrier or lane price
adjustments specifically focused on under-performing segments, typically resulting in short-term
fixed-price agreements). With index-based strategies, the contracted price between a shipper and
primary carrier increases or decreases with a market-based indicator such as those reported by
Cass Information Systems Inc. or DAT Freight & Analytics.
1The Cass Truckload Linehaul Index is comprised of 95% contract load transaction data and 5% spot load data.
Discussion of how our empirical dataset appropriately represents dynamics in the truckload freight market is found
in Section 5
Acocella, Caplice, and Sheffi: Opportunities for market-based freight contracts
Article submitted to ; Please do not circulate! 7
A handful of large shippers that operate in the consumer packaged goods, food and beverage, and
manufacturing industries, and their carriers (both asset and non-asset) have explored index-based
pricing strategies in practice. Those that have to date, however, have only done so on a small set
of trial or pilot lanes (Sinha and Thykandi 2019). In addition, some carriers, particularly 3PLs and
brokerages (i.e., non-asset providers), have dedicated resources to offer dynamic pricing options for
their shipper customers (see Convoy (2020), Schneider Transportation (2019), Uber Freight (2021),
and BR Williams (2020)).
Despite the potential benefits of market-based pricing strategies, the advantages of traditional
fixed-price contracts remain. Shippers and carriers alike seek consistency and predictability - both
in terms of demand or capacity availability and prices. Shippers run annual RFPs to establish fixed
contract prices around which they can budget. Carriers intend to provide high-quality service to
their customers but need to know they will be able to cover their costs in the process. Keeping
this in mind, we aim to help shippers and carriers understand where this traditional fixed-price
approach is effective and where an alternative market-driven approach may be beneficial.
3 Literature Review
In this research, we quantify carrier’s willingness to stick to contracts by constructing empirical
behavioral models that include industry dynamics not captured in previous literature.
3.1 Supply chain contracts
Much of the extant literature on supply chain contracting explores the ways contracts help coordi-
nate or share risks between buyer and supplier (Lariviere 1999). Agents seek risk-sharing contracts
to encourage both sides to remain committed to the contract terms through buy-back (Pasternack
1985, Deneckere et al. 1997), revenue sharing (Cachon and Lariviere 2005), and options (Barnes-
Schuster et al. 2002) contracts. (See Cachon (2003) and Simchi-Levi et al. (2014) for surveys of
these contracting mechanisms.)
The non-binding TL freight contract is similar to the self-enforcing agreements studied in the
context of repeated games literature, which explores agreements that result in both parties main-
taining the terms of the agreement over time without the interference of an external party (Telser
1980). Instead, informal mechanisms are used for contract enforcement. As the transportation
provider considers each load, if it sticks to the contract by accepting the load, it maintains that
relationship with its shipper. On the other hand, the transportation provider can defect from the
contract, reject the load, and damage the relationship with each rejection (see Scott et al. (2020)).
This self-interested, opportunistic behavior that violates the existing contract may be influenced
to higher priced alternative options on an external spot market (Williamson 1975, 1985, Wathne
Acocella, Caplice, and Sheffi: Opportunities for market-based freight contracts
8Article submitted to ; Please do not circulate!
and Heide 2000). However, the value of future potential business - i.e., the shadow of the future -
encourages carrier contract compliance (Heide and Miner 1992, Oxley 2012).
For shippers, the incentive to stick to their contracts stems from the initial investment and vetting
involved with securing the contracted carriers in the first place. The procurement process discussed
in Section 1.1 takes months of preparation, communication, and analysis to establish contracted
carriers on each lane. Transaction Cost Theory considers the costs incurred as a result of the infor-
mation search, negotiation, and monitoring involved in the exchange of a good or service between
agents – effectively, the frictions of a transaction (Williamson 1985, 1979). The theory justifies the
use of contracts as a means of defining the terms of inter-firm agreements (Williamson 2002), made
particularly important with uncertainty and high frequency of interactions (Williamson 1979).
Masten (2010) underscores the benefit of contract use, noting that the contract collects many
repeated heterogeneous transactions under one common pricing policy, reducing the transaction
costs needed to repeatedly negotiate prices for each transaction.
3.2 Shipper-carrier relationships
We further motivate the research by expanding the extant body of literature on shipper-carrier
relationships, specifically that on primary carrier load acceptance decisions. Rather than direct
measures of the shipper-carrier interactions, however, much of the literature considers attributes
of the lanes and freight to determine primary carrier acceptance rate (PAR). PAR is measured as
the percentage of loads a primary carrier accepts relative to the number of loads it is tendered on
its contracted lanes. High lane volume (Harding 2005), low lane volume volatility (Kim 2013), high
prices (Amiryan and Bhattacharjee 2015), and high lane consistency, or cadence, (Aemireddy and
Yuan 2019) have been found to be positively correlated with higher PAR, all else equal.
The shipper-carrier relationship itself, however, has been found to contribute to carrier PAR in a
few studies. Scott et al. (2016) analyze contract and spot market transactions and the impact of the
history of the shipper-carrier relationship on freight acceptance. The authors find that less frequent
load offers increase the likelihood of a primary carrier rejecting a load, while higher offered volume,
lower load offer volatility, and higher revenue transacted between the shipper and carrier increase
the likelihood of carrier’s load acceptance. Zsidisin et al. (2007) define a “good” shipper-carrier
relationship as one in which a contract is in place and find that contracted carriers outperform non-
contracted carriers in terms of freight acceptance, on-time delivery, and pre-positioned capacity.
These two studies each obtain data from a single shipper, which limits their ability to generalize
across types of shippers or segment their datasets to offer insights specific for types of shippers’
Acocella, Caplice, and Sheffi: Opportunities for market-based freight contracts
Article submitted to ; Please do not circulate! 9
The impact of market dynamics on carriers’ freight acceptance decision is studied in Acocella
et al. (2020). The authors consider two distinct market conditions: soft, where shippers have low
demand relative to available capacity and thus corresponding low contract and spot prices, and
tight, where demand for TL capacity outstrips supply and prices are high (spot prices typically
exceed contract prices). The authors find one of the main contributor to primary carriers’ tight
market acceptance decision to be how competitive the shippers’ contracted load offer price is with
prevailing market prices. Scott et al. (2016) also consider a measure of market condition - Spot
Premium - in determining carrier load acceptance decisions and find that carriers consider overall
market conditions and alternative priced loads when making contracted load acceptance decisions.
The nature of existing TL contracts, external market conditions, and their impact on carriers’
load acceptance decisions are studied in Scott et al. (2020). The authors find that explicit contracts
(those that have more formally defined service level expectations) as compared to implicit contracts
illicit higher primary carrier acceptance rates from carriers. In addition, and similar to the findings
by Acocella et al. (2020), as the market tightens and becomes more attractive to carriers, the
benefit seen by these explicit contracts diminishes.
Finally, Lindsey and Mahmassani (2015) consider TL carriers’ reservation prices and willingness
to accept loads. Using spot market prices and a discrete choice survey experiment, the authors
find that carriers’ contracted load acceptance decisions are impacted by higher-priced hypothetical
alternative load options. We expand this research by using empirical data to study actual carrier
decisions and explore how these decisions differ across shippers’ freight networks, carriers’ service
types, and overall freight market conditions.
3.3 Supplier price stickiness
While our research builds on the transportation literature, we also draw from that which models
price stickiness - in other words, the rate at which prices change in response to internal firm or
external market dynamics. Much of this literature focuses on how suppliers’ prices change as a
collective at the industry level, rather than the micro, firm level.
The most relevant stream of literature to our research consists of a set of studies focused on
producer pricing stickiness to customer demand patterns or to external market prices. Manufac-
turers adjust prices due to underlying costs and customer demand changes (Loupias and Sevestre
2013) and beliefs that competitors are also changing prices (known as coordination failure) (Blinder
(1991) and Blinder et al. (1998)).
This coordination failure comes into play in the TL context during the strategic RFP stage.
Carriers do not coordinate with one another on pricing of lanes. However, while carriers submit
lane bid prices according to their own network fit, if they want to be competitive, they must also
Acocella, Caplice, and Sheffi: Opportunities for market-based freight contracts
10 Article submitted to ; Please do not circulate!
factor in their beliefs about how other carriers may be pricing the lanes they want. Submitting
bids that are too high may mean another carrier is awarded the business; too low and they may
be chosen for the contract, but at a price that cuts into critical profit margins. This coordination
failure may result in a race to the bottom and cause the shipper to award poor-fitting carriers to
Studies by Fabiani et al. (2006) and Fabiani et al. (2007) use a survey of over 11,000 firms across
Europe and find that most firms consider both historical market information and expectations
of the market when making pricing decisions. The authors conclude that not only is price stick-
iness influenced by market conditions, it is also related to customer relationships as defined by
explicit and implicit contracts. These findings underscore truckload practitioner sentiments and
findings by Scott et al. (2020) that a combination of market conditions and standing shipper-carrier
relationships are key determinants for carrier pricing and load acceptance decisions.
The above studies also find that individual firms’ rate of price change tends to be slower than
that of the general market. This time- and market state-dependent supplier price stickiness is
further explored by Loupias and Ricart (2006), Hall et al. (2000), and Apel et al. (2005). Similar
results are observed in the TL freight context as carriers’ contract price changes tend to lag those
of spot prices (see Pickett (2018)). In addition, contract freight rates rise more quickly than they
The extant literature shines light on how producers make (predominantly strategic) pricing
decisions. However, there are few studies of if or how suppliers choose to offer their product or
service at an operational level to contracted customers or to non-contracted customers on a spot-
like market. We aim to add to the literature in this way.
4 Research Question and Hypotheses
As discussed above, the long-term, fixed-price contracts between shippers and their TL carriers are
non-binding in volume and capacity commitments. They often result in degraded performance and
unexpected price escalations as exogenous market conditions change. Alternative contract forms,
in particular those based on indexed pricing, have drawn recent attention. With this research, we
aim to help shippers and carriers determine the most promising areas within their networks for
index-based pricing in their portfolio of freight contracts. Thus, we address the following research
For which segments of shippers’ networks - lanes, volume, demand patterns, and
carriers - should they consider index-based contracts?
To answer this question, we model primary carrier load acceptance decisions based on the load’s
offered price relative to the carrier’s best-known alternative option: the current lane-specific spot
Acocella, Caplice, and Sheffi: Opportunities for market-based freight contracts
Article submitted to ; Please do not circulate! 11
market price. In this way, we can demonstrate carriers’ contract price stickiness. We formulate our
first hypothesis as:
H1: A load is more likely to be accepted by a primary carrier the higher the contracted price is
relative to the current, lane-specific spot market price.
This claim is the basis for our behavioral assumptions that carriers prefer higher priced loads
(Williamson 1975, Wathne and Heide 2000).
Next, we consider carriers’ willingness to stick with contract prices for different lane and freight
types. We measure how primary carriers’ likelihood of accepting a load is impacted as the offered
price of the load changes relative to the spot market price for these lane and freight segments of
interest. First, we consider a lane’s distance. While carriers each have individual preferences and
strategies relating lane distances, we aim to isolate segments to determine general patterns. We
define the following hypothesis:
H2: The likelihood a primary carrier accepts a load increases with lane distance and further
increases as the load’s contracted price increases relative to the current, lane-specific spot price.
Next, we consider how price adjustments impact primary carrier load acceptance for different lane
demand patterns. Shippers’ infrequent and inconsistent tendering behaviors lead to lower primary
carrier acceptance (see Section 3.2). For lanes on which a shipper tenders loads inconsistently, it
may be more difficult or require longer distances driven empty for a carrier to re-position capacity
when a load is tendered. A higher price incentive may improve the probability the carrier accepts
the load.
We expect that index-based pricing for lanes with inconsistent demand will result in higher
primary carrier acceptance and less reliance on backup or spot alternatives. Moreover, carriers
would benefit by being able to better cover additional internal costs for serving unanticipated
demand. This leads us to the following two hypotheses:
H3: The likelihood a primary carrier accepts a load increases with higher shipper tendering fre-
quency on the awarded lane and further increases as the load’s contracted price increases relative
to the current, lane-specific spot price.
H4: The likelihood a primary carrier accepts a load increases with lower shipper tendered volume
volatility on the awarded lane and further increases as the load’s contracted price increases relative
to the current, lane-specific spot price.
In the above hypotheses, we capture lane-level demand patterns that make carrier acceptance
difficult. Next, we measure a load characteristic: surge volume. Recall, the contract agreement
between a shipper and a primary carrier includes a (non-binding) awarded or expected volume
However, due to poor forecasts, network changes, or other unanticipated end customer demand,
Acocella, Caplice, and Sheffi: Opportunities for market-based freight contracts
12 Article submitted to ; Please do not circulate!
shippers may tender more loads to a contracted carrier than the awarded volume in a given week.2
This additional surge volume above the carrier’s allocated capacity to that lane may be more
difficult or costly to cover and may require an additional price incentive for the carrier to accept
the surge loads. Thus, we identify each load that is tendered above the awarded weekly volume as
surge loads and test the following hypothesis:
H5: The likelihood a primary carrier accepts a surge load increases as the load’s contracted price
increases relative to the current lane-specific spot price.
Finally, we aim to determine which carrier types are best suited for indexed pricing. Carriers
fall into one of two categories based on the services provided: asset-based carriers that own the
trucks and trailers used to move freight, and non-asset providers, often referred to as brokers or
third party logistics (3PL) providers. This latter group of providers remove the shippers’ burden
of securing capacity and act as the middle man between shippers and asset-based carriers. Brokers
access a vast pool of typically smaller, asset-based carriers, aggregate their capacity, and match it
to shippers’ demand. These brokers, or non-asset providers, typically buy and sell transportation
based on their expectations of the general market; their profit margins are tied to how well they
are able to manage the cycles. We test the following hypothesis:
H6: Non-asset primary carriers are less likely than asset-based primary carriers to stick to fixed-
price contracts as spot market prices rise.
We test this hypothesis by calculating the rate of primary carrier load acceptance (PAR) as Spot
Rate Differential changes. A higher rate of change (i.e., decrease in PAR as spot price increases
relative to contract price) suggests lower contract price stickiness. A similar approach is taken by
Lindsey and Mahmassani (2015) to measure carriers’ load price elasticity.
We further characterize asset carriers based on their fleet size. We expect that larger carriers may
be better insulated from fluctuations in the market and may be more willing to forego opportunistic
spot market loads than smaller carriers. We formulate our final hypothesis as follows:
H7: Larger asset carriers are more likely than smaller carriers to stick to fixed-price contracts
as spot market prices rise.
By exploring each of these hypotheses individually, we can measure carriers’ contract price
stickiness for different freight segments of interest and identify which areas of a shipper’s network
each carrier type may be most amenable to index-based pricing. In this way, we address our central
research question.
2A similar issue for primary carriers may arise if the shipper tenders much less than the awarded volume. While this
is an important issue to address, it is out of the scope of the present research.
Acocella, Caplice, and Sheffi: Opportunities for market-based freight contracts
Article submitted to ; Please do not circulate! 13
5 Carrier load acceptance model specification
In this section, we summarize our empirical data, describe our carrier acceptance model, and define
the variables we use to build the model. Our partner company - a major US freight transportation
management firm - provides us with transaction data over four years (2015-2019) of all of the TL
loads for 68 shippers of various sizes and industry verticals and their 412 (primary and backup)
carriers, both asset and non-asset. The data represent each load’s tender sequence. This includes
the date, time, and price at which each load is tendered to the primary carrier with its accept
or reject decision, and, if needed, backup carriers’ accept or reject decisions. The reported tender
sequence continues down the routing guide waterfall until a carrier - primary or otherwise - accepts
the load and the price the load is accepted at, including an indication if it ultimately is moved on
the spot market.
The set contains 1.7 million long-haul loads (i.e., loads that move a distance greater than 250
miles)3all of which originate and terminate in the continental US. From the subset of tenders to
primary carriers, we model the probability a load is either accepted or rejected (a binary outcome)
by the primary carrier based on associated load, lane, shipper, and carrier characteristics described
in the following subsections.
Logistic regression models are widely used in econometric literature to isolate the relationships
between a binary dependent variable and independent input variables. Moreover, these models are
the predominant modeling choices for authors studying producer and consumer price stickiness
(Loupias and Sevestre 2013, Cecchetti 1986, Cao et al. 2012). We adopt a logistic model choice
as well and, to allow for non-linearity between independent and dependent variables (Morris and
Joyce 1988), we discretize continuous input variables and include them as categorical variables.
Our dataset is comprised of multiple load accept/reject decisions by each carrier. As such, we
must account for repeated measures of the same individual carrier. With repeated measures data,
typical (logistic) regression modeling neglects to account for the correlations between the set of
decisions made by the same individual. This within-subject correlation results in inefficient estima-
tors. In other words, the calculated estimators have a greater spread around the true population
values. Instead, General Estimating Equations (GEEs) model the average response of an individ-
ual in the population (Liang and Zeger 1986, Ballinger 2004). The regression coefficient estimates
in GEE models consider the covariance matrix between the outcomes in the sample associated
with the same individual. GEE estimators reduce to those obtained through OLS if the dependent
variable is normally distributed and no within-individual response correlations exists (Hardin and
Hilbe 2012, Greene 2003).
3We use this long-haul distinction because pricing structures for the alternative, short-haul moves, differ from those
we consider and discuss in this research.
Acocella, Caplice, and Sheffi: Opportunities for market-based freight contracts
14 Article submitted to ; Please do not circulate!
We use a GEE model with a logistic link function where the coefficient estimates represent the
marginal increase or decrease in the (log) odds a carrier accepts a load:
logit(yc,k ) = logyc,k
1yc,k =xT
where each individual primary carrier, c, makes a binary accept/reject decision for each load it
is tendered, k(the positive outcome, yc,k = 1, is a load acceptance), and the matrix of explana-
tory variables, xc,k, are the lane, freight, shipper, and carrier variables described in the following
5.1 Load Spot Rate Differential
We define the Spot Rate Differential (SRD) in Section 2 and Eq. 1 as the percent difference between
the current lane-specific spot market price and the contracted price of a load. The SRD calculation
requires knowledge of each lane’s spot price at the time a load is offered. While we do not have
such data consistently across all lanes and time, given the breadth of our dataset, we can calculate
benchmark spot market prices.
Our dataset approximately represents the general freight market trends as it is comprised of
many shippers’ load tenders across the continental US. We corroborate this claim by comparing
two statistics from our dataset to external industry data. First, we measure the correlation between
the time series of average primary carrier acceptance rate (PAR) in our dataset, a real-time pri-
mary carrier behavior, and the Morgan Stanley Freight Index, which represents overall practitioner
sentiment of the market’s supply and demand. The correlation between the two time series is 85.2%
(see Scott et al. (2016) for a similar justification process). Second, the correlation between our
national average contract linehaul price and that of the Cass Truckload Linehaul Index is 91.8%.
We conclude that our dataset is sufficiently representative of the overall freight market.
Next, we reconstruct spot prices for each origin region to destination region combination. Regions
of the US differ in attractiveness to carriers based on the business opportunities that are expected
in those areas. For example, regions with high outbound demand are typically more attractive for
carriers to accept loads going into because they are more likely to easily find a follow-on load. To
account for the dynamic nature of spot prices, we include the year and month indicator in the spot
price model.
Daganzo (2005) shows that point-to-point transportation costs (e.g., linehual costs) result from
a combination of fixed and variable costs. Based on this, we model spot prices using multiple linear
regression with heteroskedastic robust standard errors. We regress the point-to-point linehaul price
of loads that are accepted on the spot market in our dataset on origin and destination region
Acocella, Caplice, and Sheffi: Opportunities for market-based freight contracts
Article submitted to ; Please do not circulate! 15
binary variables and a month and a year binary variable. These represent fixed costs. We include
a continuous distance variable, which represents the carriers’ variable costs (Ballou 1991, Scott
2015). Acocella et al. (2020) further detail the price benchmarking methodology used here.
Our base case lane (i.e., the binary variables omitted to avoid multicollinearity) originates in
the Lower Atlantic region of the US and terminates in the South Central region in January of
2016. These ommitted variable choices correspond to the most volume (i.e., greatest number of
observations) within each categorical variable. The lane-specific spot price for a given time, Spoti,j,t
is defined as:
Spoti,j,t =ˆ
βbase +ˆ
βdistX(i,j )
dist +X
The intercept term, ˆ
βbase, is the fixed cost of the base case, and ˆ
βdist, is a distance (i.e. variable
cost) coefficient associated with the average distance of loads between iand j,X(i,j)
dist . The fixed
cost of the origin and destination regions that are different from the base case are represented
by the I-1 origin coefficients, ˆ
βi(where Iis the set of origin regions and Xithe corresponding I
binary variables indicating in which region the load originates), and J-1 destination coefficients, ˆ
(where Jis the set of destination regions and Xjthe corresponding destination binary variables).
The 15 mutually exclusive and collectively exhaustive regions are key market areas defined by our
industry partner representing geographic clusters of transportation demand patterns. The origin
and destination coefficients of our linear regression model, ˆ
βiand ˆ
βj, can be interpreted as spot
price premiums associated with an origin or destination different from the base case lane.
Finally, ˆ
βmand ˆ
βymeasure the dynamic, time-based changes in spot prices for each month
and year, respectively, and capture both seasonal and underlying market structural trends.4As a
notational simplification, for the remainder of this paper, we combine mand yfor a time-dependent
variable by denoting it with a subscript t.
With the results of this model, we calculate a single average spot price for every origin-
destination-month-year combination in the dataset. As discussed in Section 1.2, this spot price
represents the average of a distribution of underlying spot load prices. We use this average spot
market price to calculate the Spot Rate Differential (SRD) for each load, k, using Equation 1 as
the key metric to test carriers’ contract price stickiness.
4We choose to include month and year variables (as opposed to a single variable that treats each month in the time
frame separately - that is, m∈ {1,60}and no Xyvariable) to ensure enough observations are in each time-based
outcome variable for robust coefficient estimates.
Acocella, Caplice, and Sheffi: Opportunities for market-based freight contracts
16 Article submitted to ; Please do not circulate!
5.2 Lane distance
Carriers’ preferences regarding lane distance may vary depending on individual operations. Each
carrier tries to maximize asset utilization - in other words, miles driven carrying a paid load. We
measure a lane’s distance by the number of days it takes to drive that distance, assuming a driver
can drive 400 miles per day based on federally mandated hours of service regulations and actual
driving patterns (J.B. Hunt 2015). For a more detailed discussion of carrier economics, see Belzer
and Sedo (2018), Masten (2010), and Burks and Monaco (2019).
Our lane distance measure is broken into five binary “travel-days” variables based on whether
the lane is expected to take up to 1 day to drive, 1-2 days, 2-3 days, 3-4 days, or more than 4 days.
Rather than using a continuous variable for distance (or travel-days), we decompose the variable
into these categories to avoid a linear assumption on the relationship between carrier acceptance
and distance.
5.3 Lane demand cadence
In the supplier-buyer relationship, frequency of interactions points to more positive relationship;
infrequent demand patterns are problematic for suppliers (Rinehart et al. 2004). This aspect of
the relationship is particularly true in the TL industry. Inconsistent or infrequent tendered volume
from one shipper (the customer) makes it difficult for the transportation suppliers (carriers) to
plan where and when they need to position capacity to balance their networks and serve each of
their other customers.
One metric carriers use to measure shipper performance is tender cadence, or the frequency at
which loads are tendered (J.B. Hunt 2015, C.H. Robinson 2015). Scott et al. (2016) incorporate
this frequency of interactions as a contributor to load acceptance by measuring the number of
days since the previous load was offered to a carrier by a shipper. The authors find that carriers
are less willing to accept loads that are tendered at unpredictable frequencies. Moreover, Acocella
et al. (2020) demonstrate that primary carrier acceptance increases with the percentage of weeks
in which loads are tendered to that carrier on that lane.
We use this latter measure to characterize a lane’s frequency. For each load, we calculate the
percent of the preceding four weeks in which that shipper tendered at least one load to that carrier
on that lane. The resulting cadence metric is a discrete measure, taking on values of 0%, 25%, 50%,
75%, or 100%. Our input variables to the carrier acceptance decision models are binary variables
indicating which of these discrete values the cadence measure takes.
Acocella, Caplice, and Sheffi: Opportunities for market-based freight contracts
Article submitted to ; Please do not circulate! 17
5.4 Lane demand volatility
In addition to the frequency of interactions, consistency of demand is an important factor for
carriers to anticipate capacity needs (J.B. Hunt 2015, C.H. Robinson 2015). Both practitioners
and the literature note the importance of reducing tendered volume variability to improve carrier
freight acceptance, reduce the cost of loads, and help carriers better utilize their capacity (Harding
2005, Scott et al. 2016, Aemireddy and Yuan 2019, Acocella et al. 2020).
We incorporate a measure of lane-level tendering volatility between a shipper and primary car-
rier acceptance. Each load is assigned the corresponding lane-level tendering volatility, which we
calculate as the shifted 4-week rolling average week-over-week change (measured as the square
difference) in tendered volume from the shipper to the carrier on that lane:
V oli,j,t =sPt4
τ=t1(di,j,τ di,j,τ 1)2
where load kis tendered on lane (i, j) in time period tand di,j,τ is the number of loads tendered
on the lane in week τ. We consider only weeks in which loads materialize and are tendered to
the carrier - in other words, weeks in which di,j,τ >0. This is because we want a measure of the
volatility of the materialized volume. We already capture how frequently there are no-volume weeks
with the Cadence measure.
5.5 Surge volume
Next, we consider surge volume, a load characteristic defined as tendered loads that are above
the awarded, or expected, weekly volume. Carriers report that they can typically manage to make
capacity available when the number of loads tendered from a shipper in a week is within about
10% of the lane award volume. However, as volume reaches and surpasses 20% of the award
volume, carriers often are unable to serve the excess demand. Moreover, shippers commonly call for
contracted carriers to flex up with increased demand, often up to 20% above the awarded volume,
as a stipulation of the service level expectations in the contract (Singh 2021). Such service level
expectations highlight the non-binding nature of the contract: while agreed upon and defined in the
contract, they are not legally or contractually enforceable by the shipper. The main incentive for
the carrier to uphold them is the promise of continued business from the shipper (i.e., the shadow
of the future).
While the expected volume is part of the information communicated to carriers during the RFP,
many shippers do not keep careful record of the awarded volume to each carrier on each lane after
the bid is complete. As such, our dataset does not include the primary carriers’ awarded volume
for each lane. Instead, we use a proxy for this awarded volume: the preceding shifted 4-week rolling
Acocella, Caplice, and Sheffi: Opportunities for market-based freight contracts
18 Article submitted to ; Please do not circulate!
average of the tendered volume to the primary carrier on the lane. Scott et al. (2016) use a similar
proxy to rank loads. The authors measure the average daily volume on a given lane over the 30
days preceding the load of interest and denote how the load’s rank within the day measures relative
to that 4-week rolling average daily volume. We similarly rank each load within the week and
categorize it based on how its rank compares to the awarded weekly volume proxy.
Each load is assigned to its corresponding surge category. If the load’s rank within the week is
less than or equal to the awarded volume proxy, it is given a Surge category of Within Mean. If
the load’s rank is more than the average but less than or equal to 10% above the mean, it is in
the Up to 10% Surge category, if it is more than 10% but less than or equal to 20% of the average
volume, it is in the Up to 20% Surge category, and if the load’s rank is more than 20% above
the awarded volume proxy, it is in the Over 20% Surge category. We expect the higher the surge
volume category a load is in, the less likely it is to be accepted by a primary carrier.
5.6 Carrier service type
Asset and non-asset carriers offer different services which drive their relationships with shippers and
how they interact with the overall market. On one hand, asset carriers own the trucks (tractors) they
operate. They have fixed available capacity to manage across their networks and serve customers.
On the other hand, non-asset carriers - otherwise known as brokers or freight forwarders - do
not own the tractors or trailers that move their customers’ loads. Instead, non-asset providers
match shippers with asset-based carriers. The advantage to shippers of working with a non-asset
carrier is that they are able to aggregate smaller (asset) carriers’ capacity to serve the shipper’s
needs. Often, shippers aims to maintain a manageable carrier base size (i.e., number of contracted
carriers) across their networks. Rather than contracting with many, small carriers, they prefer to
allow brokerages to manage these carriers. The benefit for smaller asset carriers of working with a
broker or 3PL is that often they might not otherwise have access to some shippers’ business. For
larger asset carriers, the broker may provide opportunities to fill backhauls or moves to re-position
a truck that does not have contracted volume and would otherwise be an empty, unpaid trip.
A shipper may set up a contract with a non-asset provider on a lane to lower the risk that loads
go to the less predictable, volatile spot market. Brokers set prices with shippers for a contract
term length and then typically pay (close to) spot market prices for their asset carriers over the
course of the contract. They aim to hedge the market and, over time, still make a reasonably
sustainable margin. When spot market prices are high, they may be paying carriers more than
they are receiving from their contracted shippers. However when spot market prices are lower in a
soft market, they receive higher payments from their fixed-price contracts than the price at which
they are buying capacity on the market.
Acocella, Caplice, and Sheffi: Opportunities for market-based freight contracts
Article submitted to ; Please do not circulate! 19
Because of the difference between how asset and non-asset providers utilize markets, we predict
non-asset carriers are more likely than asset carriers to be pulled from their contracted loads when
spot prices are high and more attractive than contract prices. Thus they would be more responsive
to index-based pricing.
Using an auction theory lens, this distinction in behaviors between how asset and non-asset
carriers approach bid pricing is addressed in Scott (2018). We aim to expand on this by modeling
the behaviors of asset and non-asset carriers separately.5In this way, we extract each type of
carriers’ contract price stickiness for different freight and lane segments independently.
5.7 Asset carrier fleet size
Within the asset provider segment, carriers’ cost structures vary by fleet size. For example, large
carriers may be more able to absorb variations in market prices than would a smaller, owner-
operator that owns a single truck and trailer. The distribution of asset carrier fleet size - both
across the industry in the US and in our dataset - is highly skewed; about 60% of total for-hire
carriers in the US are independent owner-operators, and 96% of fleets have fewer than 20 trucks.
To account for this, we include the log of the carrier’s fleet size (tractor count) as our measure of
asset carrier size.
Figure 3: Distribution of Asset Carrier Fleet Size Figure 3 depicts the
skewed distribution of
carriers’ number of trac-
tors in the fleet and the
normalized distribution
of the log of carrier fleet
size from our dataset.
5.8 Other fixed effects
While we do not make for-
mal hypotheses regarding
the following variables, we do control for their fixed effects in our model and report their effects
on carrier acceptance decisions. These include the shipper’s size, measured by the log of its total
monthly tendered volume across all lanes, and the shipper’s industry vertical.
5Some carriers also offer both services. For example, large asset carriers J.B. Hunt, Schneider, and Knight-Swift all
run a brokerage division of their business. In our dataset, we can distinguish between whether a company’s asset or
non-asset arm was awarded or tendered loads because they fall under different SCAC codes. SCACs, or Standard
Carrier Alpha Codes, are unique 2- or 4-letter codes assigned by The National Motor Freight Traffic Association, Inc.,
(NMFTA) to identify transportation companies and are used throughout the freight industry for consistent carrier
Acocella, Caplice, and Sheffi: Opportunities for market-based freight contracts
20 Article submitted to ; Please do not circulate!
Indicators for the lane’s origin and destination regions are also included to account for their
relative attractiveness. As discussed in Section 5.1 and Acocella et al. (2020), carrier acceptance is
expected to vary across different inbound or outbound regions due to the business opportunities
present at the origin and destination. We include in our carrier acceptance model a binary variable
for each of the 15 regions of the US defined by general market demand patterns of our transportation
management partner company, at both an origin and destination.
The results of our model indicate which load, lane, and carrier characteristics are statistically
significant indicators for predicting primary carrier acceptance rate (PAR) at a load transaction
level. Further, we address our carrier price stickiness research question by quantifying the changes
in PAR as the contracted load price relative to the spot market price (i.e. SRD) changes for the
resulting statistically significant freight and carrier segments.
5.9 Market Condition
Building off the research outlined in Section 3, we expect carriers’ contract price stickiness to differ
depending on the general market condition. Our dataset spans two distinct market conditions: a
soft market observed from the first week of February 2016 to the first week of July 2017 and again
after the second week of January 2019 until the end of the time frame covered by the dataset;
and a tight market observed before the first week of February 2016 and from the first week of
July 2017 to the second week of January 2019. Acocella et al. (2020) justify these market periods
by identifying the weeks in which a statistically significant change is observed in the underlying
structure of the TL freight market. We develop four distinct acceptance models: (1) asset carriers
in soft markets, (2) non-asset carriers in soft markets, (3) asset carriers in tight markets, and (4)
non-asset carriers in tight markets.
6 Results
In this section, we summarize our results by discussing the full GEE logistic regression models
and use them to predict the (log) likelihood a load is accepted for each hypothesis. The regression
results are tabulated in the electronic companion. The separate reported tables comprise a single
model with the input variables described in Sections 5.1-5.8.
As validation of each of our four models, we report the Brier Score, which is the appropriate
scoring metric when probability outcomes are the desired result (Wallace and Dahabreh 2014,
Niculescu-Mizil and Caruana 2005, Zadrozny and Elkan 2002, Brier 1950). The Brier Score, BS ,
is the mean square error between the predicted probability an observation is in the the positive
class, pn(here, an accepted load, yc,k = 1), and the actual outcome, on∈ {0,1}:
Acocella, Caplice, and Sheffi: Opportunities for market-based freight contracts
Article submitted to ; Please do not circulate! 21
BS =1
where Nis the total number of observations in the dataset. The Brier Score takes on values
between 0 and 1. Better models have lower Brier Scores.
To build our models, we segment each of the datasets for the four models into a training set on
which we fit the model and a test set to measure model performance with a 70%:30% split. In each
of the four pre-split datasets, the relative frequency of accepted loads is much higher than that of
rejected loads. To develop unbiased models that do not naively favor accepted load predictions, we
use a stratified sampling technique to define each of the training and test sets such that the ratio
of the number of accepted loads to number of rejected loads in each split set is the same as that
of the original pre-split dataset. In this way, the validated models can better predict carriers’ load
acceptance or rejection probabilities on new data.
Table 1: Test set Brier Scores
Model Brier Score
Asset carriers
Soft market 0.049
Asset carriers
Tight market 0.032
Non-asset carriers
Soft market 0.075
Non-asset carriers
Tight market 0.064
The Brier Scores for each of our four models on their respec-
tive test datasets are reported in Table 1. Recall that the
closer a score is to 0, the better. A “good” Brier Score largely
depends on the dataset itself. However, we draw from the lit-
erature for reasonable comparison. Zadrozny and Elkan (2002)
report test set Brier Scores for the best calibrated probabil-
ity prediction models of five datasets ranging from 0.012 to
0.204. Similarly, Wallace and Dahabreh (2014) report calibrated model Brier Scores for 34 datasets
between 0.042 to 0.319. All four of our models’ Brier Scores sit below the average Brier Score value
of the best fitting models in these previous studies. Thus, we conclude that our models perform
well in predicting loads acceptance probability.
6.1 Carrier price stickiness by service type and market condition
The results of the GEE logistic regression models are reported in Table EC.1. Hypothesis H1,
which addresses carriers’ overall contract price stickiness is supported: in general, primary carriers
are more likely to accept a load with a contract price that is higher relative to the current, lane-
specific spot market price. The statistically significant coefficients in Table EC.1 show that as SRD
increases and the spot price rises above contract prices, the probability a carrier accepts the load
decreases for both carrier types in both market conditions. This suggests that in general, carriers
can be incentivized away from their contracted loads as the external spot market increases relative
to those contract prices.
Figure 4 illustrates this behavior for each carrier type and market condition. To construct the
figure, we plot the probability a load is accepted at each of the statistically significant SRD values
Acocella, Caplice, and Sheffi: Opportunities for market-based freight contracts
22 Article submitted to ; Please do not circulate!
for a base case lane and fit a linear model to these plotted results./footnoteWe show only the
segments of these linear models over the SRD ranges that apply to the associated market condition.
In other words, soft market models apply when SRD negative (spot market prices are below contract
prices) and tight market models apply when SRD is positive. The slope term of each linear model
represents the carrier’s contract price stickiness, or willingness to stick to contracted loads as spot
market prices change relative to contract prices in the corresponding market condition.
Both carrier types are about twice as likely to be pulled from contract load prices in a tight
market as they are in a soft market: asset carriers have a slope of -0.29 in tight markets as compared
to -0.12 in a soft market and non-asset carriers have a slope of -0.65 in soft markets and -1.32 in
tight markets. Moreover, H7 is supported: non-asset carriers are about five times more likely to be
pulled from contracted loads than their asset-based counterparts in each market condition.
For shippers considering indexed contract pricing strategies, these results suggest that all else
equal, non-asset carriers are more likely to respond to market-based pricing than asset carriers and
that both carrier types may be more responsive to such pricing strategies in tight markets than
Figure 4 Carrier Contract Price Stickiness Models
Next, we discuss the results for specific freight and lane segments highlighted in our hypotheses.
We demonstrate carriers’ contract price stickiness by holding spot price constant at $1000 for both
soft and tight markets and plotting the contract price needed to maintain a 90% likelihood the
primary carrier accepts the contracted load over the range of values each freight or lane segment of
interest takes. We choose 90% PAR because most shippers expect at least this level of service from
Acocella, Caplice, and Sheffi: Opportunities for market-based freight contracts
Article submitted to ; Please do not circulate! 23
their contracted carriers. Many may expect higher acceptance rates, however this threshold repre-
sents basic service level expectations in the shipper-carrier relationship. Our method of presenting
contract price stickiness follows those discussed in Section 3 on general supplier price stickiness
and responses to factors exogenous to the supply contract.6
6.2 Contract price stickiness: lane demand consistency
Results for both lane tendering frequency and volatility are summarized in EC.3. They show that
hypotheses H3 and H4 are supported particularly for asset carriers in both market conditions.
Carriers prefer lanes with more frequently and consistently tendered volume; that is, carriers are
more willing to stick to their contract prices on lanes with high frequency load tendering and lower
week-to-week volume volatility.
Importantly, the shipper often cannot control its load tendering consistency; it is subject to
external factors such as its own customers’ demand patterns, inbound suppliers’ schedules, and
congestion on roadways or at ports. While the shipper does not control when its demand for trucks
materializes, it does control what carriers are offered the loads when they do appear. Moreover,
the shipper has historical knowledge of its lane demand patterns. Thus, the shipper’s decision here
is what bid prices to accept and how to tender loads that do appear for lanes with historically
irregular cadence or inconsistent volumes.
6.2.1 Tendering cadence
The results weakly support Hypothesis H3: lanes on which loads are tendered less frequently see
lower primary carrier acceptance. Moreover, both asset and non-asset carriers in both market
conditions are less willing to stick to their contract prices for these low cadence lanes. (See EC.3.)
Figure 5 shows that as lane tendering cadence decreases, asset carriers’ contract price needed to
maintain 90% PAR increases about 7% in soft markets (to $682 per load from $617) and 2.5% in
tight markets (to $857 from $830).
Similarly, non-asset carriers require a 6% contract price increase in soft markets to maintain
90% PAR on low cadence lanes. While they do not appear to make load acceptance decisions
based on lane tendering cadence in tight markets, these non-asset carriers set the highest contract
prices for all cadence levels in tight markets. This may be because shippers often use non-asset
brokerage services specifically for infrequently tendered lanes. These providers may be familiar
with the undesirable, infrequently tendered lanes and knowingly set higher contract prices during
the RFP.
6The figures identify segments where primary carriers require higher price incentive to stick to their contracts as
the spot price approaches or surpasses the contract price. Contract prices of categorical variable values for which
the GEE regression model coefficients are not statistically significant (see EC.1) are presented as the contract price
associated with the baseline value.
Acocella, Caplice, and Sheffi: Opportunities for market-based freight contracts
24 Article submitted to ; Please do not circulate!
Figure 5: Contract Price Stickiness, Tender Cadence
The results suggest that both
carrier types are willing to stick
with contract prices on moderate
and high cadence lanes under both
market conditions. However, lanes
with very infrequent load tenders
require higher contract prices rel-
ative to going spot market prices.
Shippers may see better primary
carrier acceptance on these low fre-
quency lanes with market-based pricing strategies. Important to note, transportation practitioners
and the literature both highlight that there must be enough business between the buyer and sup-
plier (on a lane) to offset the additional effort involved in introducing either non-traditional, or
more explicit contracts (Scott et al. 2020, Sinha and Thykandi 2019).
6.2.2 Tendered volume volatility
Next, we consider the tendered volume volatility on a lane from a shipper to its primary carrier.
Consistent with Hypothesis H4, higher tendered volatility leads to lower load acceptance probabili-
ties, particularly for asset carriers. The GEE logistic regression model results are presented in Table
EC.3 and Figure 6 below demonstrates carrier contract price stickiness. In soft markets, asset carri-
ers stick to their contract prices for lanes with week-to-week tendered volume volatility up to 50%.
Figure 6: Contract Price Stickiness, Lane Volatility
However, as tendering
volatility increase, they are
more likely to be pulled from
the contract priced loads
unless contract prices are
much closer to the current
spot market price. Specifically,
contract prices needed to
maintain 90% acceptance are
17% higher on these high volatility lanes than low volatility lanes in soft markets.
In tight markets, as compared to soft markets, even higher contract prices are needed for asset
carriers to maintain high acceptance rates. Moreover, these price escalations begin at lower volatility
lanes (i.e., 11-25%). Contract prices needed on high volatility lanes are 26% higher than those on
Acocella, Caplice, and Sheffi: Opportunities for market-based freight contracts
Article submitted to ; Please do not circulate! 25
low volatility lanes in tight markets. This suggests that in tight markets, asset carriers are more
easily pulled from their contracts on lanes with lower tendering volatility than in soft markets.
Non-asset carriers are more willing to stick to their contract prices at high lane volatility in
both market conditions as compared to their asset-based counterparts. Contract prices needed for
non-asset providers to maintain 90% load acceptance rates remain steady for lanes with more than
10% week-over-week change in tendered volume, with an increase of only 5% in soft markets for
moderate and high volatility lanes.
Shippers should pay close attention to moderate and high volatility lanes in their networks.
These lanes are particularly difficult for asset carriers to accommodate due to fixed capacities
and a network of many customers’ demand they continuously balance while non-asset providers
do not. For the lanes on which a shipper’s demand volatility is difficult to control or smooth out
(by splitting the volume and tendering to multiple carriers, for example), it may be beneficial to
introduce a market-based pricing strategy that adjusts contract load prices as spot market prices
change to better incentivize primary asset carriers to accept loads.
6.3 Contract price stickiness: surge volume
Table EC.4 summarizes the GEE logistic regression models’ results demonstrating carriers’
likelihood of accepting loads considered surge volume. Hypothesis H5 is strongly supported
for asset carriers: asset primary carriers are less likely to accept loads that are above the
lane awarded volume - particularly those over 20% above awarded volume - than loads
within their expected weekly tendered volume. Put another way, asset carriers are less will-
ing to stick to their contract prices for excessive surge volume, especially in tight markets.
Figure 7: Contract Price Stickiness, Surge Volume Figure 7 shows that the con-
tract price required to maintain
90% acceptance for asset carriers
in soft markets increases by 9% for
loads that are over 20% above the
awarded volume. These asset car-
rier contract prices further increase
in tight markets. For surge volume
over 20% above the award in tight
markets, shippers pay $940 per load
- almost at the $1000 spot market price level - which equates to a 13% increase in contract price
from the awarded volume loads.
Acocella, Caplice, and Sheffi: Opportunities for market-based freight contracts
26 Article submitted to ; Please do not circulate!
The contract price required to maintain high acceptance rates for loads up to 10% and 10-20%
above awarded volumes with asset carriers stays relatively steady. This moderate surge volume
requires contract price increases of 4-6% in both soft and tight markets. Asset carriers are willing
to stick to their contract prices for these mid-level surge loads. This may be because shippers
often communicate an expectation of primary carriers to accept anywhere from 10-20% above
the awarded volume without reducing acceptance or other carrier performance metrics in their
(non-binding) contract service level agreements. Although asset carriers may have more difficulty
providing this additional capacity than non-asset carriers, they may have factored some additional
capacity into their strategic capacity allocation decisions during the original RFP. Surge loads over
this 20% threshold require higher contract prices, especially in tight markets, and asset carriers
are more willing to be pulled from contracts for better priced alternatives on the spot market.
Non-asset carriers, on the other hand, do not appear to be incentivized away from contracts
with higher priced alternatives on surge volume, even for loads that are more than 20% above
the awarded volume. This difference between asset and non-asset carriers is expected. Non-asset
carriers are not limited in capacity in the way asset carriers are. While non-asset carriers must also
balance supply and demand of capacity, they are better equipped to serve excess demand because
they can access a large pool of asset carriers and aggregate capacity accordingly. Moreover, non-
asset carriers already require higher contract prices to maintain 90% acceptance for all volume - in
fact, very close to the spot market price - regardless of surge classification. Thus, they may already
be adequately incentivized to offer that capacity.
Shippers may see the most benefit from implementing indexed pricing applied to specific surge
volume loads with asset carriers. This type of volume-based pricing strategy has been seen in
practice. It is referred to as tier pricing, where levels of surge volume are set at a higher fixed price
- determined during the strategic RFP - than that of the base awarded volume.
6.4 Contract price stickiness: lane distance
Table EC.2 summarizes the lane distance portion of the GEE logistic regression model results. It
shows weak support of Hypothesis H2: non-asset carriers prefer longer lanes. In soft markets, loads
on shorter lanes - specifically those with only a one-day travel time - are less likely to be accepted
than all other lane distances. In tight markets, loads on longer lanes (three- and four-day lanes) are
more likely to be accepted than loads on shorter lanes. However, there is no statistically significant
difference in probability of load acceptance for asset carriers on different lane distances, suggesting
that asset carriers do not consider lane distance in their load acceptance decision.
Acocella, Caplice, and Sheffi: Opportunities for market-based freight contracts
Article submitted to ; Please do not circulate! 27
6.5 Asset carrier contract price stickiness by fleet size
The preceding sections consider carriers based on their service type: asset and non-asset. However,
asset carriers range widely in their fleet sizes (Figure 3), which directly impact their internal
cost structures, tolerance to price fluctuations, access to network model optimization software to
plan strategically, and as a result, willingness and ability to accept loads. Table EC.5 summarizes
the GEE logistic regression models’ results for an asset carriers’ fleet size. Hypothesis H8 is not
supported: all else equal, we do no find statistically significant evidence that the size of the asset
carrier (measured here as log of Tractor Count) indicates its primary acceptance rate in soft or
tight markets.
7 Discussion and Implications for Index-based Pricing
In this study we explore supplier price stickiness in response to customer demand patterns and
exogenous market conditions. We take the case of firms that outsource their TL transportation
service needs and aim to help them determine when, where, and with which transportation suppliers
they should consider alternative pricing strategies from the standard, fixed-price contracts. In
particular, we consider dynamic market-based pricing by measuring primary carriers’ contract
price stickiness, or the change in contract price relative to the going spot price needed so carriers
maintain a reasonable load acceptance service level.
1. Shippers should not delay or skip RFPs or mini-bids as markets tighten. Both
carrier types (asset and non-asset) are twice as likely to stick to their contract priced loads in
a soft market than a tight market. Shippers can use this to inform the timing of their strategic
procurement event as market conditions may be changing. For example, say a shipper runs its
annual strategic RFP in January. In the preparation months leading up to the event, the shipper
may observe that spot market prices or other leading indicators of general market conditions are
rising. Some shippers in this situation may consider delaying the RFP and keeping the current
contracted prices that had been set during a softer, lower-priced market rather than opening itself
up for carriers to lock in higher prices. Alternatively, the shipper may choose to run the RFP
anyway, perhaps allowing for slightly increased contract prices, and assume it can expect good
primary carrier acceptance rates for the next year even as markets further tighten. Finally, the
shipper may take a middle-ground approach and proactively increase rates for certain core carriers
on specific important lanes to keep them out of an RFP.
However, the results of our study suggest that no matter the decision, if spot prices continue
to rise as the market tightens further, the contract prices will become less competitive (i.e., their
spot rate differential will increase) and primary carriers with low contract price stickiness will
begin opportunistically rejecting loads at higher rates, defecting to either fresher contract prices
Acocella, Caplice, and Sheffi: Opportunities for market-based freight contracts
28 Article submitted to ; Please do not circulate!
with other shippers or to the spot market. Therefore, when markets are tightening, shippers need
to continuously ensure their contract prices stay competitive with the going spot market prices,
regardless of their RFP timing. One approach shippers can take as market conditions become more
constrained is to execute smaller, focused, more frequent “mini-bids” with core primary carriers
on the most important, susceptible lanes. These mini-bids result in shorter term (e.g., 30-, 60-,
or 90-day) contracts. Another approach is to implement a market-based pricing strategy, as we
propose in this study, which effectively ensures contract prices are up-to-date.
2. Non-asset carriers are best suited for market-based contracts. All else equal, asset
primary carriers are five times more likely to stick to their contract priced loads than non-asset
primary carriers, or brokers, in both soft and tight markets. In fact, across most network and
freight segments, shippers must pay these non-asset primary carriers just about spot market prices
to maintain high load acceptance rates. This suggests that brokers may respond to index-based
contracts better than asset-based providers.
3. Market-based pricing shows promise for volatile demand, low frequency lanes,
and surge volume. Asset primary carriers are less likely to stick to their contract priced loads
on lanes with infrequently tendered loads and high week-to-week volatility of tendered volume.
Shippers often have little control over when loads actually materialize and carriers’ capacity is
needed. However, they do know historical demand patterns on their lanes. They can control which
carriers they tender the loads that do materialize and the pricing strategies they are willing to
implement. Our results suggest a market-based pricing strategy for these low cadence or high
volatility lanes would better incentivize asset carriers to stick to their contracts.
In addition to these lanes types, shippers can expect asset carriers to respond to market-based
pricing for surge volume - in other words, loads that are in excess of the expected weekly volume
communicated during the RFP. This finding relates to the two stages of TL transportation. During
the first stage, the strategic RFP, the shipper communicates the expected weekly volume on each
lane that the carriers bid on. Once the carrier wins the lane, it plans to allocate capacity for
that expected volume, perhaps with an additional 10-20% buffer. It also uses this knowledge of
expected demand to balance the demand for trucks across its existing network and to inform its
bid strategy in other shippers’ RFPs. However, during the second stage, the operational stage
when loads materialize and are tendered to that carrier, the actual demand may well exceed the
capacity for which the carrier has planned. Asset carriers in particular are constrained in their
total capacity available at any one time. To maintain high acceptance rates of loads that exceed
20% over the awarded volume on a lane, carriers require a price incentive. This is an important
segment of shippers’ freight: consistently across our time horizon, 10-15% of total tendered volume
is in this surge category.
Acocella, Caplice, and Sheffi: Opportunities for market-based freight contracts
Article submitted to ; Please do not circulate! 29
In fact, a tier pricing strategy where surge volume is priced higher than the baseline awarded
volume is used by some shippers and their larger or core carriers, precisely for the reasons described.
However, these tier-based rates are still fixed at the time the contract is established. A dynamic
market-based price, on the other hand, would ensure the tiered price incentive remains competitive
with the current market.
4. Indexed pricing may result in lower load acceptance if applied in soft markets.
Important to note that we discuss market-based pricing with the implication that contracted load
prices increase as market prices increase in a tight market, but also decrease as prices decrease in
soft markets - in other words, “symmetric” indexing. Shippers may want to consider using dynamic
market pricing as an incentive for primary carriers whereby indexed contract prices only increase
as the market tightens, but settles to the competitive, fixed contract price when the market prices
are decreasing. Otherwise, for segments that require a contract price premium (i.e., higher contract
price than spot market price) for high primary carrier acceptance, the shipper would expect to see
decreased primary carrier acceptance when the indexed price symmetrically decreases.
Our study suggests that there is still a place for the widely used fixed-price contracts. On lanes
where tendered volume is consistent and frequent, both shippers and carriers prefer a set price
to plan and budget around. While these prices may also become outdated as spot market prices
increase, carriers are more willing to stick to their contracts for attractive freight.7Moreover,
competitive fixed price contracts may still be best suited for soft market conditions.
8 Limitations and Future Research
Our empirical modeling results offer both academic and practical contributions. First, we add to
the econometric literature on supplier price stickiness in response to demand and exogenous market
dynamics. We do so by quantifying TL transportation suppliers’ price stickiness as market prices
change and for different customer demand patterns, freight segments, characteristics of shippers’
networks, and supplier service types. Our work explores real-time decision that previous literature
overlooks; transportation suppliers have the spot market as a dynamic alternative option to their
contracted business for every transaction.
In addition, we add to the literature specific to shipper and carrier relationships, particularly
with a focus on how changing market conditions impact behaviors. We do so by utilizing a uniquely
extensive and detailed dataset. Previous empirical studies in the space have been limited in that
their data represent only a single shipper’s business, and often contain limited or no tendering or
carrier acceptance service level information.
7While low tight market period primary carrier acceptance rates are undesirable, the majority of freight still moves
under fixed contract rates even in tight markets. However, we find evidence that a market-based approach would
improve primary carrier acceptance rates.
Acocella, Caplice, and Sheffi: Opportunities for market-based freight contracts
30 Article submitted to ; Please do not circulate!
This study is not without its limitations. Precisely due to the observational nature of our dataset,
we cannot control for outside influences on carriers’ acceptance decisions. We attempt to account
for these factors by segmenting the contract price stickiness analysis by lane, load, and carrier
types and other fixed effects that have been previously identified as contributing factors. In doing
so, we build models with good performance (i.e., very low Brier Scores). However, there may still
be alternative explanations for carriers’ acceptance decisions.
Notwithstanding these limitations, our study further adds to the existing shipper-carrier relation-
ship literature by isolating the impact of market prices on the way carriers make freight acceptance
decision for their contracted shippers. As discussed in the motivation of this research, there has
been growing interest in index-based pricing in the TL industry. This research serves as a starting
point. We demonstrate where shippers can expect to see carrier acceptance behaviors most influ-
enced by dynamic, market-based pricing methods. An interesting future stream of literature could
develop strategies for shippers and carriers to design and implement these freight contracts.
The authors would like to thank the anonymous reviewers for greatly improving the quality of this manuscript.
Acocella, Angela, Chris Caplice, Yossi Sheffi. 2020. Elephants or goldfish?: An empirical analysis of carrier
reciprocity in dynamic freight markets. Transportation Research Part E: Logistics and Transportation
Review, 142 102073.
Aemireddy, Nishitha Reddy, Xiyang Yuan. 2019. Root cause analysis and impact of unplanned procurement
on truckload transportation costs. M. Eng in Logistics Thesis, MIT.
Amiryan, Nane, Sharmistha Bhattacharjee. 2015. Relationship between price and performance: An analysis
of the us trucking market. M. Eng in Logistics Thesis, MIT.
Apel, Mikael, Richard Friberg, Kerstin Hallsten. 2005. Microfoundations of macroeconomic price adjustment:
Survey evidence from swedish firms. Journal of Money , 37 313–338.
Ballinger, Gary A. 2004. Using generalized estimating equations for longitudinal data analysis. Organizational
Research Methods, 7 (2), 127-150.
Ballou, Ronald. 1991. The accuracy in estimating truck class rates for logistical planning. Transportation
Research - Part A: Policy and Practice, 25 (6), 327-337.
Barnes-Schuster, Dawn, Yehuda Bassok, Ravi Anupindi. 2002. Coordination and flexibility in supply con-
tracts with options. Manufacturing & Service Operations Management, 4 (3), 171-207.
Belzer, Michael H., Stanley A. Sedo. 2018. Why do long distance truck drivers work extremely long hours?
Economic and Labour Relations Review, 29 (1), 59-79.
Blinder, Alan S. 1991. Why are prices sticky? preliminary results from an interview study. American
Economic Review, 81 89-96.
Blinder, Alan S., Elie Canetti, David Lebow, Jeremy Rudd. 1998. Asking about prices: A new approach to
understanding price stickiness. The Journal of Consumer Affairs, 32 (2), 424-427.
BR Williams. 2020. What is “index-based pricing” in transportation and logistics
and how can it help my organization? URL
what-is-index-based- pricing-in-transportation-and-logistics/. Accessed March 29, 2021.
Brier, Glenn W. 1950. Verification of forecasts expressed in terms of probability. U.S. Department of
Commerce Monthly Weather Review, 78 1-3.
Acocella, Caplice, and Sheffi: Opportunities for market-based freight contracts
Article submitted to ; Please do not circulate! 31
Burks, Stephen, Kristen Monaco. 2019. Is the u.s. labor market for truck drivers broken. Monthly Labor
Review, 142 (3), 1-23.
Cachon, G’erard P. 2003. Supply chain management: Design, coordination, and operation. Handbooks in
Operations Research and Management Science, 11 227-339.
Cachon, G’erard P., Martin A. Lariviere. 2005. Supply chain coordination with revenue-sharing contracts:
Strengths and limitations. Management Science, 51 (1), 30-44.
Cao, Shutao, Wei Dong, Ben Tomlin. 2012. The sensitivity of producer prices to exchange rates : Insights
from micro data. Tech. Rep. 2012-20, Bank of Canada Working Paper.
Caplice, Chris. 2007. Electronic markets for truckload transportation. Production and Operations Manage-
ment, 16 (4), 423-436.
Cecchetti, Stephen G. 1986. The frequency of price adjustment: A study of the newsstand prices of magazines.
Journal of Econometrics, 31 255–274.
C.H. Robinson. 2015. Do ‘favored shippers’ really receive better pricing and service? White Paper.
Convoy. 2020. A new approach to primary freight. URL
white-paper-guaranteed-primary/. Industry White Paper, Accessed April 1, 2021.
Daganzo, Carlos. 2005. Transportation costs. chap. 2.3. Logistics Systems Analysis. 23-29. 4th ed.
Deneckere, Raymond, Howard P. Marvel, James Peck. 1997. Demand uncertainty and price maintenance:
Markdowns as destructive competition. The American Economic Review, 87 (4), 619-641.
Fabiani, Silvia, Martine Druant, Ignacio Hernando, Claudia Kwapil, Bettina Landau, Claire Loupias, Fer-
nando Martins, Thomas Matha, Roberto Sabbatini, Harald Stahl, Ad C. J. Stokman. 2006. What
firms’ surveys tell us about price-setting behavior in the euro area. International Journal of Central
Banking, 2 3-48.
Fabiani, Silvia, Claire Suzanne Loupias, Fernando Manuel Monteiro Martins, Roberto Sabbatini, et al. 2007.
Pricing decisions in the euro area: how firms set prices and why. OUP USA.
Greene, William H. 2003. Econometric Analysis. 5th ed. Pearson Education.
Hall, Simon, Mark Walsh, Anthony Yates. 2000. Are uk companies’ prices sticky? Oxford Economic Papers,
52 425–446.
Hardin, James, Joseph Hilbe. 2012. Generalized Estimating Equations (GEE), vol. 99. Chapman & Hall.
Harding, Matthew. 2005. Can shippers and carriers benefit from more robust transportation planning
methodologies? M. Eng in Logistics Thesis, MIT.
Heide, Jan B., Anne S. Miner. 1992. The shadow of the future: Effects of anticipated interaction and
frequency of contact on buyer-seller cooperation. Academy of Management Journal, 35 (2), 265-291.
J.B. Hunt. 2015. 660 minutes: How improving driver efficiency increases capacity. Industry White Paper.
Kim, Yoo Joon. 2013. Analysis of truckload prices and rejection rates. M. Eng in Logistics Thesis, MIT.
Lariviere, Martin A. 1999. Supply chain contracting and coordination with stochastic demand. Sridhar
Tayur, Ram Ganeshan, Michael Magazine, eds., Quantitative Models for Supply Chain Management ,
chap. 8. Springer, 233-268.
Liang, Kung-Yee, Scott L. Zeger. 1986. Longitudinal data analysis using generalized linear models.
Biometrika, 73 (1), 13-22.
Lindsey, C, H S Mahmassani. 2015. Measuring carrier reservation prices for truckload capacity in the
transportation spot market choice experiment. Transportation Research Record, 2478 123-132.
Loupias, Claire, Roland Ricart. 2006. Price setting in the french manufacturing sector: New evidence from
survey data. Revue d’Economie Politique, 116 541–554.
Loupias, Claire, Patrick Sevestre. 2013. Costs, demand, and producer price changes. Review of Economics
and Statistics, 95 (1), 315-327.
Masten, Scott E. 2010. Long-term contracts and short-term commitment: Price determination for heteroge-
neous freight transactions. American Law and Economics Review, 11 (1), 79-111.
Acocella, Caplice, and Sheffi: Opportunities for market-based freight contracts
32 Article submitted to ; Please do not circulate!
Morris, Michael H., Mary L. Joyce. 1988. How marketers evaluate price sensitivity. Industrial Marketing
Management, 17 (2), 169-176.
Niculescu-Mizil, Alexandru, Rich Caruana. 2005. Predicting good probabilities with supervised learning.
Proceedings of the 22nd international conference on Machine learning. 625-632.
Oxley, Fabrice Lumineau Joanne E. 2012. Let’s work it out (or we’ll see you in court): Litigation and private
dispute resolution in vertical exchange relationships. Organization Science, 23 (3), 820-834.
Pasternack, Barry Alan. 1985. Optimal pricing and return policies for perishable commodities. Marketing
Science, 4 (2), 166-176.
Pickett, Chris. 2018. The coyote curve: A model for mitigating risk and uncertainty in modern supply chain
operations. Coyote Logistics.
Rinehart, Lloyd M, James A Eckert, Robert B Handfield, Thomas J Page, Thomas Atkin. 2004. An assess-
ment of supplier-customer relationships. Journal of Business Logistics, 25 (1), 25-62.
Schneider Transportation. 2019. Management optimizes packaging manufacturer’s spot freight
with index pricing model. URL
how-market-index-pricing- model-optimizes-spot-freight-shipping. Industry White Paper.
Scott, Alex. 2015. The value of information sharing for truckload shippers. Transportation Research Part E:
Logistics and Transportation Review, 81 203-214.
Scott, Alex. 2018. Carrier bidding behavior in truckload spot auctions. Journal of Business Logistics, 39
(4), 267–281.
Scott, Alex, Christopher Craighead, Chris Parker. 2020. Now you see it, now you don’t: Explicit contract
benefits in extralegal exchanges. Production and Operations Management, 29 (6), 1467-1486.
Scott, Alex, Chris Parker, Christopher Craighead. 2016. Service refusals in supply chains: Drivers and
deterrents of freight rejection. Transportation Science, 51 (4), 1086-1101.
Simchi-Levi, David, Xin Chen, Julien Bramel. 2014. The Logic of Logistics. 3rd ed. Springer.
Singh, Omar. 2021. Perfectly executed routing guides fail 100% of the time. URL https://www. routing-guides-fail-100-of- the-time.
Surge Transportation.
Singh, Ranjit, Ramesh Billa, Bhupinder Singh Arora. 2016. Index based pricing: Managing risk and prof-
itability. Deloitte, .
Sinha, Atmaja, Rakesh Thykandi. 2019. Alternate pricing model for transportation contracts. M. Eng in
Logistics Thesis, MIT.
Sokoloff, David, Gaohui Zhang. 2020. Predicting and planning for the future: North american truckload
transportation. M. Eng in Logistics Thesis, MIT.
Telser, Lester G. 1980. A theory of self-enforcing agreements. The Journal of Business , 53 (1), 27-44.
Uber Freight. 2021. Uber freight expands procurement solutions with market access. URL http://www. Accessed March 29, 2021.
Wallace, Bryon C., Issa J. Dahabreh. 2014. Improving class probability estimates for imbalanced data.
Knowledge and Information Systems, 41 33-52.
Wathne, Kenneth H., Jan B. Heide. 2000. Opportunism in interfirm relationships: Forms, outcomes, and
solutions. Journal of Marketing, 64 (4), 36-51.
Williamson, Oliver. 1975. Markets and Hierarchies: Analysis and Antitrust Implications. Free Press, New
Williamson, Oliver. 1985. The Economic Institutions of Capitalism: Firms, Markets, Relational Contracting.
Free Press, New York.
Williamson, Oliver E. 1979. Transaction-cost economics: The governance of contractual relations. The
Journal of Law and Economics, 22 (2), 233-262.
Williamson, Oliver E. 2002. The theory of the firm as governance structure: From choice to contract. The
Journal of Economic Perspectives, 16 (3), 171-195.
Acocella, Caplice, and Sheffi: Opportunities for market-based freight contracts
Article submitted to ; Please do not circulate! 33
Zadrozny, Bianca, Chrles Elkan. 2002. Transforming classifier scores into accurate multiclass probability
estimates. Proceedings of the ACM SIGKDD International Conference on Knowledge Discovery and
Data Mining. 694-699.
Zsidisin, George, Douglas Voss, Matt Schlosser. 2007. Shipper-carrier relationships and their effect on carrier
performance. Transportation Journal, 46 (2), 5-18.
E-Companion for “The end of ‘set it and forget it’ pricing?
Opportunities for market-based freight contracts”
This e-companion reports the detailed results of the logistic regression model, which is used to
quantify carriers’ contract price stickiness for segments indicated by model variables.
Table EC.1 GEE model results: Spot Rate Differential, asset and non-asset carriers, soft and tight markets
Spot Rate
Differential Asset carriers
Soft market Asset carriers
Tight market Non-asset carriers
Soft market Non-asset carriers
Tight market
Constant 2.0032 1.2945 0.634 1.0894
(0.882) (1.121) (0.837) (1.451)
SRD: 0.4862** 0.6254* 0.3237 0.2436
(<50%) (0.236) (0.382) (0.431) (0.466)
SRD: 0.6307*** 0.7545** 0.5789** 0.4886
[50%,45%) (0.233) (0.332) (0.292) (0.846)
SRD: 0.3642* 0.1127 0.2974 0.0262
[45%,40%) (0.202) (0.272) (0.425) (0.415)
SRD: 0.3445 0.242 0.6138** 0.0223
[40%,35%) (0.233) (0.290) (0.279) (0.432)
SRD: 0.467*** -0.0195 0.5006* -0.1933
[35%,30%) (0.183) (0.245) (0.267) (0.436)
SRD: 0.3231* -0.1767 0.4178* 0.4921
[30%,25%) (0.195) (0.233) (0.239) (0.444)
SRD: 0.1934 0.028 0.0378 0.7571***
[25%,20%) (0.196) (0.209) (0.211) (0.253)
SRD: 0.2504 -0.1155 0.0674 0.2454
[20%,15%) (0.200) (0.198) (0.218) (0.269)
SRD: 0.2193 -0.1083 0.0619 -0.2322
[15%,10%) (0.171) (0.177) (0.204) (0.406)
SRD: 0.0146 0.1091 0.1712 -0.1569
[10%,05%) (0.151) (0.179) (0.238) (0.266)
SRD: -0.0314 0.2073* -0.1154 0.3754*
[05%,0%) (0.150) (0.124) (0.256) (0.202)
SRD: omitted omitted omitted omitted
SRD: 0.0264 -0.2026 -0.0804 -0.0055
[05%,10%) (0.1360) (0.149) (0.395) (0.250)
SRD: -0.0054 -0.1423 -0.0226 -0.0968
[10%,15%) (0.165) (0.148) (0.246) (0.247)
SRD: 0.026 -0.1789 0.1061 -0.2498
[15%,20%) (0.149) (0.158) (0.309) (0.361)
SRD: -0.152 -0.1171 0.5717** -0.8346***
[20%,25%) (0.297) (0.182) (0.267) (0.340)
SRD: 0.0913 -0.2481 -0.2945 -0.5186
[25%,30%) (0.208) (0.191) (0.436) (0.385)
SRD: -0.3266 -0.1904 0.7472*** -0.5058
[30%,35%) (0.287) (0.189) (0.308) (0.371)
SRD: -0.1179 0.266 -0.0413 -0.5918**
[35%,40%) (0.300) (0.201) (0.638) (0.286)
SRD: -0.2114 -0.1257 1.8206*** -0.7198**
[40%,45%) (0.325) (0.225) (0.695) (0.314)
SRD: -0.286 0.0457 1.1126 -0.6647**
[45%,50%) (0.299) (0.217) (0.800) (0.332)
SRD: 0.0406 -0.1655 0.9579*** -0.8166***
[>50%) (0.275) (0.207) (0.320) (0.284)
Note: robust standard errors reported in parentheses
significance level: *0.1; **0.05; ***0.01
Table EC.2 GEE model results: Lane Distance (measured in Travel Days), asset and non-asset carriers, soft and
tight markets
days Asset carriers
Soft market Asset carriers
Tight market Non-asset carriers
Soft market Non-asset carriers
Tight market
1 day -0.0152 0.0594 -0.3804** 0.3756*
(0.167) (0.168) (0.192) (0.212)
2 days omitted omitted omitted omitted
3 days 0.2134 -0.0518 0.0866 0.6837***
(0.220) (0.162) (0.268) (0.206)
4 days 0.3755 -0.3351 -0.2664 0.8952**
(0.260) (0.244) (0.363) (0.409)
>4 days 0.5188 -0.0202 0.3404 0.2598
(0.402) (0.373) (0.309) (0.268)
Note: robust standard errors reported in parentheses
significance level: *0.1; **0.05; ***0.01
Table EC.3 GEE model results: Lane Tendering Consistency, asset and non-asset carriers, soft and tight markets
variable Asset carriers
Soft market Asset carriers
Tight market Non-asset carriers
Soft market Non-asset carriers
Tight market
Cadence: -0.3690*** -0.2296** -0.4965*** -0.3933
25% (0.105) (0.119) (0.195) (0.249)
Cadence: -0.1291** -0.1145 -0.1757 -0.0943
50% (0.063) (0.088) (0.141) (0.152)
Cadence: omitted omitted omitted omitted
Cadence: -0.1195 -0.0611 0.243 0.199
100% (0.093) (0.077) (0.198) (0.200)
Volatility: 0.9023*** 0.9203*** 0.6908*** 0.6318*
Up to 10% (0.243) (0.219) (0.169) (0.373)
Volatility: 0.4429*** 0.3591*** 0.2136 0.5434***
(10-25%] (0.139) (0.106) (0.219) (0.161)
Volatility: omitted omitted omitted omitted
Volatility: -0.3142*** -0.3789*** 0.0508 0.2485**
(50-75%] (0.065) (0.098) (0.132) (0.110)
Volatility: -0.5076*** -0.6222*** -0.050 0.1623
(75-100%] (0.086) (0.133) (0.177) (0.232)
Volatility: -0.8049*** -0.8206*** -0.1754 0.1213
(100-125%] (0.119) (0.133) (0.254) (0.210)
Volatility: -0.7579*** -0.8419*** 0.0278 0.1355
(125-150%] (0.116) (0.140) (0.285) (0.220)
Volatility: -0.6011*** -0.8065*** -0.0155 0.1349
(150-200%] (0.127) (0.182) (0.304) (0.238)
Volatility: -0.4359*** -0.746*** 0.1467 0.2419
Over 200% (0.143) (0.175) (0.313) (0.266)
Note: robust standard errors reported in parentheses
significance level: *0.1; **0.05; ***0.01
Table EC.4 GEE model results: Surge Volume, asset and non-asset carriers, soft and tight markets
Category Asset carriers
Soft market Asset carriers
Tight market Non-asset carriers
Soft market Non-asset carriers
Tight market
Within 0.1757** 0.1009 0.0181 0.2668*
Mean (0.089) (0.084) (0.251) (0.146)
Mean to omitted omitted omitted omitted
10% Surge
10-20% -0.0203 -0.1837** 0.1271 0.124
Surge (0.108) (0.094) (0.294) (0.150)
Over 20% -0.1559** -0.2334*** -0.1387 -0.0099
Surge (0.079) (0.083) (0.265) (0.120)
Note: robust standard errors reported in parentheses
significance level: *0.1; **0.05; ***0.01
Table EC.5 GEE model results: Carrier Fleet Size, asset carriers, soft and tight markets
Fleet size
No. tractors Asset carriers
Soft market Asset carriers
Tight market
Log Tractor Count -0.0109 -0.0088
(0.079) (0.103)
Note: robust standard errors reported in parentheses
significance level: *0.1; **0.05; ***0.01
Table EC.6 GEE model results: Shipper Fixed Effects, asset and non-asset carriers, soft and tight markets
Shipper fixed
effects variable Asset carriers
Soft market Asset carriers
Tight market Non-asset carriers
Soft market Non-asset carriers
Tight market
Shipper size 0.4033** 0.284 0.5908** 0.3054
(log monthly volume) (0.177) (0.271) (0.27) (0.406)
Vertical: 0.655*** 0.8121*** -0.3843 0.2414
Automotive (0.216) (0.261) (0.279) (0.451)
Vertical: 0.9596*** 0.9327*** -0.4351 0.1938
F&B/CPG (0.194) (0.280) (0.266) (0.439)
Vertical: omitted omitted omitted omitted
Paper & Packaging
Vertical: 0.5557 0.5942* -0.8417* 0.5001
Manufacturing (0.381) (0.336) (0.468) (0.531)
Vertical: -0.2443 0.9402 1.3812*** -0.2445
Other (0.602) (0.636) (0.467) (0.684)
Note: robust standard errors reported in parentheses
significance level: *0.1; **0.05; ***0.01
ResearchGate has not been able to resolve any citations for this publication.
Dynamic macroeconomic conditions and non-binding truckload freight contracts enable both shippers and carriers to behave opportunistically. We present an empirical analysis of carrier reciprocity in the US truckload transportation sector to demonstrate whether consistent performance and fair pricing by shippers when markets are in their favor result in maintained primary carrier tender acceptance when markets turn. The results suggest carriers have short memories: they do not remember shippers’ previous period pricing, tendering behavior, or performance when making freight acceptance decisions. However, carriers appear to be myopic and respond to shippers’ current market period behaviors, ostensibly without regard to shippers’ previous behaviors.
To enhance performance, buying firms often use supplier governance mechanisms, such as explicit contracts, to coordinate efforts. Yet, it is unclear how these governance mechanisms would perform in extralegal exchanges, whereby suppliers may renege on the contract with little or no legal recourse. Furthermore, there is a paucity of studies examining these extralegal exchanges in the presence of an external market willing to pay higher prices, thus tempting suppliers to renege on existing agreements. Accordingly, we seek insight into the effectiveness of implicit and explicit contracts juxtaposed with the evolving payoff for a supplier’s reneging. We do so with a dataset from a buying firm and its suppliers in the for‐hire trucking industry. Our results reveal that implicit contracts do, indeed, enhance performance (i.e., suppliers reject business offerings less often), even though the buying firm has no viable legal recourse for rejected business offerings. Likewise, we find that explicit contracts (i.e., providing more specificity to the supplier) further enhance performance; yet, as the external market conditions change in the supplier’s favor, more explicit contracts’ effectiveness drastically weakens. Indeed, when external market prices reach their highest, our results suggest that explicit contracts’ benefits virtually disappear. Overall, our results provide a rare look into implicit and explicit contracts’ outcomes in an extralegal exchange when suppliers’ reneging becomes more/less threatening, thereby offering insight for researchers and managers.
When a shipper urgently needs truckload service, they often utilize the spot market. But despite its importance, little is known about this market. I analyze a longitudinal data set of auctions for spot truckload service where I observe invitations for carriers to bid, whether a bid is placed, and the bid price if one is placed, and augment this with information about the bidding carriers. Drawing upon auction theory, I suggest that a carrier's a priori characteristics (size, market specialization) explain the two primary decisions made in the auctions—whether to bid and how much to bid. I find that brokers, who act as intermediaries between shippers and asset‐based carriers, bid much more frequently and higher than asset‐based carriers. Price indexes show that broker bid prices follow similar patterns, but asset‐based carrier prices do not. The results suggest that an online marketplace linking shippers directly with the thousands of asset‐based carriers could add considerable value to the for‐hire trucking industry, a development which appears to be happening.
Contracts in the for-hire trucking industry are unusual in that, although they establish prices for different services, there is typically no legally binding obligation or penalty for either party to offer or accept a load. When a load is rejected by all contract carriers, shippers must turn to the spot market, which can significantly increase supply chain costs. Because these transactions occur between private parties, data on load acceptances/rejections and contract/spot prices have not been available to academic researchers, leaving the freight rejection problem largely unexplored. We are able to examine this problem using a detailed transactional data set of a large national shipper. We estimate that spot prices for truckload services average about 62% higher than contract rates. We find key operational and economic factors to be drivers of freight rejection and the shipper-carrier relationship to be a deterrent to freight rejection. We also find that primary and secondary carriers respond differently to these operational and economic factors. We discuss how these insights could be used by a shipper to get better performance and lower cost from their carrier base.
Theoretical research undertaken over the last decades showed that the nature of nominal rigidities plays a key role in determining the effects of different shocks on the economy. This research has made clear that a thorough understanding of the extent and causes of the sluggish adjustment of nominal prices is crucial to the design and conduct of monetary policy. This book presents the main results of a research program undertaken by the Eurosystem central banks on price setting decisions by firms in the euro area. Its objective is to deepen our understanding of the behavioral mechanisms driving agents' pricing decisions, adopting a methodological approach-asking firms directly about how they set the price of their output (their pricing strategies) and why (the rationale of these strategies)-that is particularly well suited for the purpose at hand. The book also compares results for the euro area to similar analyses for other countries and summarizes the main findings of studies based on individual quantitative micro data on consumer and producer prices carried out for most euro area countries. Finally, the book explores the monetary policy implications of the main findings.
The transportation function is critical to efficient and effective logistical operations. In order to improve transportation performance, it may be necessary for firms to form closer relationships with their carriers. Unfortunately, there are few studies examining the nuances and outcomes of shipper-carrier relationships. The focus of this work is to delineate the levels of relational closeness found between shippers and carriers and to assess the effects of relational closeness on carrier performance. Based on data and expertise provided by The Hershey Company, a major U.S. confectionary corporation, this research utilizes qualitative and quantitative techniques to accomplish these tasks. Findings indicate that closer relationships between shippers and carriers have no effect on carriers' on-time performance but significantly influence carriers' willingness to commit assets to the shipper and accept loads during times of constrained transportation capacity.