Supply Chain Frontiers issue #57
Many supply chain executives believe in, and hope for, a single generally accepted truckload (TL) rate for any given lane. This is typically referred to as the market rate for a lane, and is found by taking the average or median of a sample of TL rates from a common origin and destination.
The value is not easy to obtain, but it has many applications. Logistics managers use the TL market rate when setting future transportation budgets. Supply chain planners use the number in the network design process. Knowing this value helps procurement departments to negotiate freight rates.
Despite its widespread usage, however, the commonly accepted market TL rate does not actually exist. It misrepresents the true nature of TL transportation, and is not a true market rate as such. As a result, its application can be extremely misleading.
There are three main reasons why the traditional TL rate is difficult to pin down.
First, it is hard to find comparable networks. TL networks are naturally sparse and tend to be unique to each firm. A recent analysis carried out by supply chain consulting firm Chainalytics of more than 100 shippers with over $10 billion in annual TL spend found very little overlap. For example, less than 3% of the lanes (defined as a 5 to 5 digit postal code) in the studied network for dry van truckload movements had two or more shippers hauling on them. So, the probability of finding other shippers with identical lanes is exceptionally low.
Second, there is the issue of confidentiality. Most shippers have implicit (or even explicit) confidentiality conditions with their carriers that prevent them from direct sharing of their TL rates.
The third and most important reason why these simple, lane-based market rates are hard to obtain is because they are a myth! In 20 years of analyzing and working in the trucking market, I have never found a lane where a single TL rate applies to all shippers. The market is too dynamic and the underlying cost drivers extend beyond simple geography.
Let’s take as an example a typical high-volume truckload lane from a recent set of data of shippers within the United States. The lane is 450 miles in length and has 10 shippers hauling over 3,000 loads per year on it. If we calculate the simple market rate as practiced by some firms, we would find it has a median rate of $2.93 for the last 12-month period. However, if we look at this value over time it only tells a small part of the overall story. For example, many shippers paid consistent rates throughout the year. This speaks to great routing guide compliance and operational discipline. The TL rates these “consistent” shippers paid, however, were anything but similar, and there were significant variations between the rates paid by individual shippers. Many other variations emerged from the analysis. For example, rates shifted when some shippers put their business out to bid; some shippers displayed no apparent rate discipline. There might be many reasons for rate discrepancies – but a simple, lane-based market rate does not detect them.
If the traditional approach to calculating the market rate is flawed, is there a more accurate method? The answer is yes.
A better approach is to develop a market rate for a specific shipper on a lane. This approach to modeling transportation rates is the opposite of traditional methods. Instead of simply averaging the outcomes (the lane rates), the shipper decomposes the transportation cost inputs, analyzes, isolates, and quantifies each of the individual cost drivers that in turn dictate the transportation rates.
In effect, instead of just blending variables to find the overall mythical average, this econometric approach decomposes and isolates each driver. It achieves a more accurate rate representation in two important ways.
First, it allows us to consider and price out different policies and practices. Individual shipper policies can account for more than 10% of the variability of TL rates. An econometric approach takes these differences into account so that, for example, we can determine the additional cost of delivering to a customer as opposed to an inter-plant move. We know how much multi-stop movements implicitly increase rates – over and above any accessorial or stop-off charges. We know the rate impact of increased corridor volumes. Additionally, this modeling approach lets us determine the TL rate impact of drop-and-hook versus live load/unload, 30- versus 60- versus 90-day payment terms, hazmat shipments, etc. Each shipper has its own fingerprint. Using an econometric model allows us to capture this unique profile, and determine what the market rate should be specific to that shipper’s practices, policies, and characteristics.
Second, this approach allows us to develop more robust models and make wider comparisons. By decomposing the transportation function, we have solved both the problem of sparse TL networks and rate confidentiality. An econometric-based model does not compare lane rates; it separates out the individual effects, to include the origin and destination locations. While shippers have very low lane overlap, they have very high region overlap. We exploit and leverage this by calculating the impact of loading (unloading) in different geographic regions. The model is able to accurately estimate the geographic cost impact of each 3-digit postal code region. This is the reason why an econometric-based model can estimate rates for a specific shipper on virtually any and every origin-destination pair.
The idea of a single universal, one-size-fits-all rate that is both applicable and accurate for all shippers on a given lane does not reflect the true complexity of truckload transportation. An econometric-based approach captures this complexity.
This article is based on a two-part series of blog posts written by Dr. Chris Caplice, Executive Director, MIT CTL, and published on the Supply Chain @ MIT blog. Read the two full posts here (part one and two).