Trucking company failures are a good barometer for general trucking market and economic health, with trucking failures at quite low levels both on an absolute and relative basis over the past several years (outside of a small increase in 2013-2014) due to a healthy backdrop for the trucking industry. In fact, truck failures reached an all-time low in 2018 across the 30 years of data FreightWaves analyzed going back to 1990 (EXIT.USA). It has been 11 years since the trucking market has seen a really significant bout of failures.
However, with spot rates down meaningfully (20 percent year-over-year and 30 percent off of peak), contract rates moving lower (and at risk of future downward revisions), costs running high from a robust 2018 leading to wage increases and excess capacity entering the market, and the potential for a squeeze due to higher diesel prices from IMO 2020, FreightWaves believes this tenuous setup could portend the first meaningful increase in truck failures in years.
SONAR Ticker: EXIT.USA
What are the leading causes of truck failures?
When contemplating the leading causes for trucking failures, the principal qualitative reasons often cited by the failing companies and the media include the following (which are not mutually exclusive and often co-dependent):
- Falling contract and spot prices
- Rising diesel prices
- Expensive, rising and unaffordable insurance costs
- High labor and maintenance costs
- Unionized labor forces
- Onerous contracts with large shippers (e.g. Amazon)
The top three statistically predictive variables
Each of the reasons listed above make a great deal of sense on the surface and likely do play some part in most failures. However, despite this long list of viable reasons, we found just three that were highly statistically predictive in our study with Michigan State University logistics professor Jason Miller. These were (in order of importance): the number of trucking failures in the prior quarter, the natural logarithm of diesel prices, and the natural logarithm of the Producer Price Index (PPI) for long-haul trucking, with diesel prices and the truckload PPI being equally important. The reason we use the natural log is that it equalizes for percentage changes, which can be key in analyzing time-series data. One important point worth noting is that the number of trucking failures in the prior quarter is admittedly a broad, catch-all measure that can and does incorporate unmeasurable negative secular industry shifts that are difficult to explicitly quantify.
FreightWaves notes that there are other factors outside the top three on the list that are important considerations. Some could be statistically predictive but the data was simply not available to run a regression on (i.e. insurance costs). Others like higher labor costs (40 percent of a carrier’s total costs) are undoubtedly a crucial expense line item given the trucking industry’s low operating margins, but were simply just not as predictive. The most probable reason for the former is though wages represent a huge expense, they are largely a fixed, visible and predictable expense. This makes a big difference when running a business.
The single most predictive variable was surprisingly the number of failures in the quarter before. Failures tend to cluster, spike episodically and persist for a number of quarters when conditions in the industry deteriorate.
The next predictive variable was the year-over-year change in the natural log of diesel prices. We use the natural log because it translates into percentage changes. The easiest way to describe why the natural log is the superior measure is that a 20-cent rise in the price of diesel today (with average diesel prices north of $3 per gallon) is not equivalent to a 20-cent rise back in the 1990s when the average price per gallon was hovering around $1. From a budgeting standpoint, we would contend that percentage changes are more impactful. This is logical given that there is a lag effect of higher diesel prices filtering into higher contract rates. Further supporting the use of the natural logarithm transformation, a model estimated using the year-over-year change in nominal diesel prices was less predictive of failures than using the year-over-year change in the natural logarithm of diesel prices.
Equally as predictive of the change in trucking failures is the year-over-year change in the natural logarithm of the truckload (TL) producer price index (PPI) for truckload, long-haul, general freight trucking (source: St. Louis Fed; U.S. Bureau of Labor Statistics). As detailed by Miller in a 2018 paper, the PPI data primarily captures TL carriers’ contract rates given the vast majority of TL freight moves under contract. We use contract rates because carriers with more than five trucks normally source between two-thirds to three-fourths of their business from the contract market. Therefore, violent moves up or down in spot rate can certainly be harmful but are more indicative of a leading indicator for where future contract rates are headed. Contract rates are not nearly as volatile as spot rates for trucking. Going back to 1997, the year-over-year changes in contract rates varied between a 10 percent increase in boom times (2018) and a 10 percent decrease in dreadful times (2009).
Lastly, there is a high correlation (0.73) between diesel prices and contract rates because as diesel prices move higher, they get passed on as fuel surcharges (at least by mid- to large-size carriers). The failure data shows that these fuel surcharges sometimes simply are not enough and do not fully pass on the diesel cost increases.
Primary conclusions from the data
The biggest surprise to most outside of trucking industry professionals would be that failures are not primarily caused by spot and contract rates falling through the floor in a recession. Instead, failures are primarily due to huge spikes in diesel prices that smaller carriers then cannot pass on.
Trucking failures are the result of large spikes in diesel prices that can’t be passed on
The chart below is very informative for a discussion on truck failures. As a note, all-time series on this figure represent year-over-year changes on a percentage (diesel and contract rates) or absolute (failures) basis.
Source: St. Louis Fed, U.S. Energy Information Administration, Broughton Capital
First, we would note that diesel prices are inherently far more volatile than contract trucking rates. It would not be atypical for diesel prices to rise or fall by 20 percent (or much more) in any given year. As previously noted, contract rates, on the other hand, vary between plus or minus 10 percent.
Second, most large spikes in truck failures (in black) are the result of a large increase in diesel prices that then cannot be adequately passed on in the form of higher fuel surcharges. This is precisely what occurred in all the major failure years in the data set: 2000-2001, 2005 and 2008-2009. This suggests that fuel surcharge programs may not be working as effectively as intended by offsetting higher diesel prices.
The average fleet size of failures over the historical data set is approximately 20 trucks, varying between 45 in 2008 and below 10 in 2005. An average fleet size of 20 trucks would be classified as a small carrier. Smaller fleets do not have the same ability to pass on fuel surcharges as large fleets; therefore, a 20 to 50 percent increase in an input cost that averages 18 percent of revenue (source: Engage) is enough to wipe out many smaller players given that operating margins for the trucking industry average only 5 percent.
What lies ahead: stepping outside the model to predict future failures
The failure model as constructed by MSU Associate Professor Miller and FreightWaves is not intended to be utilized primarily as a forecasting tool. Rather, it is designed to try and explain what factors account for the time series variance in trucking failures. That being said, we would expect a significant pick-up in failures due to the confluence of several negative factors: the lagged effect of extremely down spot prices (down 20 percent year-over-year and 30 percent off peak) will begin to filter through to lower contract rates; IMO 2020 should keep diesel prices at least flat (and potentially cause materially higher diesel prices in FreightWaves’ view); and labor/operational costs should remain high as many trucking companies increased driver pay that will not be clawed back. This dual revenue and cost pressure is likely to cause a meaningful increase in failures.
A final negative factor that is likely to come into play and cause increasing failures throughout 2019 is excess capacity with outbound tender rejections (OTRI.USA) at 12-month lows. Tender rejections are a measure of carrier optionality and with much more capacity in the market competing for lower freight volumes (OTVIY.USA), which are down 7.62% year-over-year, carriers are experiencing downward rate pressure exacerbating their increasing lack of viability.
With record levels of Class 8 new truck orders in 2018 and delivery dates well into this summer, a large number of trucks will be hitting the market just as demand begins softening and lapping hard comparisons from the prior year. This type of environment characterized by growing excess capacity is highly conducive to growing truck failures.
SONAR Tickers: OTRI.USA, EXIT.USA
FreightWaves would stop short of predicting a failure environment similar to 2008 and 2001, but a generally rising failure trend between the record low rate of failures in 2018 (310 failures) and 2001’s record high of 3,990 seems reasonable in light of deteriorating industry conditions. A failure number in the range of approximately 1,500 to 2,000 would produce the most failures in the trucking industry since 2008 (and 2005 before that). Using reasonable values for where diesel prices and contract rates will be in the second and third quarters of 2019, the model suggests failures will indeed be higher in the next two quarters relative to where they were in 2018.