Why do carriers use owner operators in the U.S. for-hire trucking industry?
Using the theoretical lenses of transaction cost economics and the structure-conduct-performance paradigm, this article examines the rationale for U.S. for-hire trucking carriers' use of owner operator contract drivers. The analysis employs a cross-sectional financial and operational data set from the U.S. Department of Transportation on the largest U.S. for-hire trucking carriers. Regression results show that the use of specialized equipment and the engagement in less-than-truckload services reduce the use of owner operators. These findings provide a strong support for Williamson's core notion of asset specificity as a driver of make-or-buy decisions. Results also demonstrate that higher segment concentration levels reduce the use of owner operators. This suggests that lack of segment competition may facilitate a higher level of vertical integration among for-hire trucking carriers as evidenced by a greater reliance on employee drivers as opposed to contract owner operators.
In the highly competitive environment of the U.S. trucking industry, for-hire carriers must decide whether to make or buy certain components or services as part of their overall strategy to maximize profit and/or minimize costs. One significant cost element for motor carriers is driver services. There are essentially two options available to for-hire carriers in acquiring these services. One is to hire employee drivers, while the second is to engage the services of independent contractors or owner operators, who own/lease their own trucks and/or trailers, while performing driving services on behalf of the for-hire carriers. Owner operators are self-employed truck drivers (Peoples and Peteraf 1999). According to the U.S. Owner Operator Independent Driver's Association (OOIDA 2008), "Owner operators include individuals or companies who own or operate one or more vehicles used to haul freight in commerce and who qualify as small businesses as defined by the regulations implemented under the authority of the Small Business Investment Act of 1958." In contrast to truck carrier employees, "Owner-operator truck drivers are prohibited under antitrust laws from forming collective-bargaining organizations" (Peoples and Talley 2007). This article focuses on the decision by for-hire motor carrier managers regarding the mix of owner operators and employees in providing driving services. We develop a model to enhance our understanding of this basic make-or-buy decision. While both for-hire and private motor carriers can employ the services of owner operators, this article focuses on the use of owner operators by for-hire carriers. A significant number of for-hire motor carriers have detailed financial and operating performance data available that facilitate the development of an explanatory model of owner operator usage. No comparable data are available regarding private carrier trucking operations.
There appears to be no single consensus strategy for owner operator use among for-hire motor carriers. Indeed, the pattern of owner operator usage varies considerably among the for-hire carriers across the various industry operating segments. The U.S. Department of Transportation last reported financial and operating performance information for large, for-hire motor carriers for the calendar year 2003. In that year, the U.S. Department of Transportation reports included financial and operating performance data for 2,137 large for-hire trucking carriers operating in the United States, 1,408 of which were engaged in general freight transportation, while 729 provided specialized freight services. Of the total general freight carriers, 69 percent used owner operators to varying degrees, with average expenses of owner operators accounting for 30 percent of the total operating expenses for these carriers, while 31 percent of the general freight carriers did not use any owner operators at all. Among the specialized carriers, 62 percent used owner operators to some extent, with owner operator expenses equaling an average of 29 percent of total operating expenses for these carriers, while 38 percent of the specialized carriers did not use any owner operators at all. Therefore, a systematic evaluation and analysis of owner operator use provides valuable insights into this important question.
There are several theoretical lenses that are applicable in accounting for the underlying rationale regarding the use of owner operators. One perspective is transaction cost economics (Williamson 1975 and 1985), which will be explained in more detail shortly. According to this view, market transactions (e.g., hiring owner operators) may be less expensive than the alternative of hiring drivers as company employees under the conditions of low uncertainty and low asset specificity. However, transaction costs may not be the only concern motivating this managerial decision, since many trucking companies employ a mix of company drivers along with owner operator drivers (Nickerson and Silverman 2003). Non-cost-related factors affecting the use of owner operators have been studied together with transaction cost factors in order to fully understand the phenomenon. He and Nickerson (2006) argued that the concern about the propriety of shipper information may discourage for-hire carriers from using owner operators lest owner operators, frequently driving for multiple carriers, pass along customer information to competing carriers. Nickerson and Silverman (2003) found that externalities on less-than-truckload (LTL) hauling and carrier stake in reputation may favor the use of company drivers over owner operators. Baker and Hubbard (2004) argued that owner operators are more common in the case of long and unidirectional shipments where back-shipments are not required. They also reported that monitoring technologies, such as on-board computers, provide an incentive for carriers to use more company drivers instead of owner operators since carriers can install monitoring technologies on their own equipment to monitor the behaviors of their employee drivers.
Trucks are not as "generic," as Klein et al. (1978) indicated three decades ago. Various kinds of specialized trucks are used to transport specialized freight, resulting in a thinner resale market and, hence, a higher stake in owning such assets. Networks of hubs and terminals facilitate more efficient movement of freight and demand more coordination effort. Furthermore, a gap in previous studies is that structural constraints have not been explicitly considered together with those firm-centric variables. It is reasonable to assume that a firm's total cost minimization effort needs to be constrained by the conditions of the market segment in which a firm is operating. The structure-conduct-performance paradigm is a useful theoretical lens to complement the transaction cost economics framework in explaining the outsourcing decision of for-hire trucking carriers. To the best of our knowledge, there has been no systematic examination of firm-specific and industry-specific variables regarding the owner operator use decision.
Grounded in the transaction cost economics framework and the structure-conduct-performance paradigm, this research adds to the literature through a simultaneous investigation of firm and industry structural variables, using the most recently available data on for-hire trucking carriers provided by the U.S. Department of Transportation. We believe that such a systematic examination of firm resources and external environment reveals a more compelling picture about how carriers make outsourcing decisions. The remainder of this article proceeds as follows: we first introduce the theoretical lenses of transaction cost economics and the structural-conduct-performance paradigm, then develop research hypotheses and present the methodology. In the subsequent sections, we provide and discuss descriptive statistics and regression results. We conclude with contributions, limitations, and further research steps.
This research combines the elements of transaction cost economics and the structural-conduct-performance paradigm to explain the use of owner operators by for-hire carriers. Transaction cost economics is a fundamental theory used to explain a firm's make-or-buy decision: the use of owner operators can be thought of as a "buy" example. In a challenge to neoclassical economics, which assumes that the market is frictionless with the price mechanism being the "invisible hand," transaction cost economics argues that market transactions incur costs, primarily due to "bounded rationality" and "opportunistic behavior" (Williamson 1985, 20). For example, managers are limited to the extent that information is available and can be processed and are likely to take advantage of others due to inherent human nature. Therefore, the firm (a form of vertical integration or hierarchy) may be a more efficient structure in order to internalize transaction costs (Coase 1937, reprinted in 1991, 20). Williamson (1985) identified two types of market transaction costs: ex ante costs, the costs of negotiating contracts, and ex post costs, the costs of monitoring performance and settling disputes when parties deviate from the contract. In the presence of transaction costs, the preferred governance structure (hierarchy vs. market) is largely determined by the degree of asset specificity and environmental uncertainty. Asset specificity may be caused by proximity and immobility of a physical site (e.g., an oil refinery), difficulty in redeploying an asset for other uses (e.g., an industrial mold for making distinctive die-castings), nontransferable skills or experience (job specific training and knowledge), and dedication (such as capacity expansion to meet a particular customer's increased demand) (Williamson 2005). Uncertainty refers to the instability occurring in a contractual relationship whenever one party is trying to change contracts without consent from the counterparty. This type of uncertainty is often caused by "random acts of nature and unpredictable changes in consumer preferences" (Koopmans 1957, 162-163), or arises from "strategic non-disclosure, disguise, or distortion of information" (Williamson 1985, 57). High levels of asset specificity and uncertainty increase transaction costs through markets and hence favor making in-house over buying through markets, and vice versa.
The structure-conduct-performance paradigm has been a dominant theoretical lens for analyzing firm strategies, especially in the field of industrial organization. Waldman and Jensen (2006) presented a good summary of this paradigm, stating that basic conditions determine market structure, market structure determines firm conduct, and firm conduct determines firm performance. In the meantime, they acknowledged that there can be feedback effects of performance and conduct on market structure. The essential assumption here is that firms in a less competitive industry (e.g., monopoly or oligopoly) are able to charge higher prices (greater than marginal costs) and obtain positive economic profits over a period of time. Primary measures of market structure include market concentration, number of competitors, and barriers to entry. Indicators of firm conduct refer to firm decision variables, such as pricing, market entry decision, product quality level, and the make-or-buy decision. The structure-conduct-performance paradigm provides a complement to transaction cost economics in accounting for firm behaviors because industry constraint influences strategic decisions such as the use of owner operators by for-hire trucking carriers.
Even though most trucks and other equipment are mobile and re-deployable in general, specialized freight carriers must use specialized equipment to handle deliveries. According to the U.S. Census Bureau (2002), specialized freight trucking industry is "primarily engaged in the transportation of freight which, because of size, weight, shape, or other inherent characteristics, requires specialized equipment, such as flatbeds, tankers, or refrigerated trailer." Particularly, in the U.S. trucking industry, specialized freight requires special handling and/ or specialized equipment, including building materials, bulk commodities, household goods, heavy machinery, motor vehicles, packages, liquid bulk in tank, and refrigerated materials. Specialized equipment can be considered freight-specific equipment and is not easily re-deployable for other uses. For example, motor vehicle trailers are not appropriate for transporting non-vehicle goods: and it is not cost efficient to transport general goods using a refrigerated truck. The initial investment in specialized equipment is usually more expensive compared to general trucking equipment. Carriers and owner operators invest in specialized equipment with the expectation that they are able to charge premiums to compensate for the additional equipment costs. Therefore, the return of such investments depends on a longer period of operating time in handling specialized freight as anticipated by investors. This dependence creates a relationship-specific risk for specialized equipment owners. Meanwhile, the demand for special freight transportation is generally smaller and more variable than it is for general freight services. On one hand, for-hire trucking carriers may find it more difficult to identify and hire a pool of qualified and available drivers who own specialized equipment. On the other hand, in case of an owner operator's failure to provide specialized freight services, for-hire carriers may not be able to find substitute labor and equipment to get the job done in a short time period and with a reasonable cost.
Previous studies have examined the owner operator issue from the perspective of asset ownership in the context of the contractual relationship between for-hire trucking firms and owner operators. Baker and Hubbard (2004) study how contractual incompleteness impacts a driver's decision to own specialized trucking assets and hence affects the availability of owner operators in the market. This study finds that the installation of on-board computers to trucks may enable carriers to overcome contracting incompleteness and hence hire more company drivers. Arrunada et al. (2004) study the use of owner operators in the European trucking industry. They argue that contracting is more difficult to achieve when specialized equipment is needed for delivery primarily due to two factors: on one hand, owner operators are less likely to own specialized equipment because the potential market may be smaller compared to the standardized equipment; on the other hand, for-hire carriers are more likely to own specialized equipment and hence use their own drivers because they are not willing to pay premiums to owner operators who own specialized equipment or they are not concerned about the hold-up problem. Therefore, since transaction costs are high due to those contingencies, for-hire carriers are better off owning equipment and using company drivers in this case. Here we propose the following hypothesis:
H1: Carriers engaged in specialized freight transportation rely less on the use of owner operators than do carriers engaged in general freight transportation.
Providing LTL services demands significant fixed investments in hubs, warehouses, break-bulk terminals, and information technologies, such as computer networks and GPS. Also, coordination among trucks and drivers is more challenging than it is in the case of truckload (TL) deliveries because one leg of delivery has an immediate impact on the next leg as well as a ripple effect on the whole LTL network. In recent years, many LTL firms have provided money-back-guarantee services for a promised delivery time window, while the promised transit times have been significantly reduced due to consumer demand for real-time tracking capabilities and supply chain efficiency. Abrupt rerouting may be very common and hence require real-time coordination with LTL drivers. A general notion is that the adoption of real-time technologies is more likely to happen among company drivers than it is among owner operators. Large LTL carriers can achieve economies of scale by installing more advanced information technologies onto company-owned trucks to centralize vehicle tracking and maintenance activities. However, owner operators may not be willing to pay for expensive technologies that are highly specific to a LTL carrier.
He and Nickerson (2006) included a LTL dummy in their regression model to explain whether a shipment is outsourced. They argued that LTL shipments, which bundle multiple shipments into single shipments, require more coordination and hence are more profitable than are TL services. Their models suggested that a LTL shipment is less likely to be outsourced compared to a TL shipment. Nickerson and Silverman (2003) provided two explanations for the argument that LTL reduced the likelihood to be outsourced: (1) Operating on a hub-and-spoke network, LTL shipments demand more coordination, which is easier to achieve by using company drivers than it is by using owner operators; (2) Goods are usually transported through multiple hands in a LTL network. Use of company drivers enables carriers to monitor delivery performance and verify theft and damage, while it is costlier and more time consuming to track down who is responsible for damage if the damaged goods were handled by multiple owner operators. Their empirical model supported the prediction that a carrier is more likely to employ company drivers than owner operators when it conducts more LTL business.
In our current study, a for-hire carrier is identified as a LTL carrier if at least half of its revenue is obtained through LTL services. In their filings to the U.S. Department of Transportation, carriers are required to report in what business they are primarily engaged and revenue breakdowns among different segments. Here we propose the following hypothesis:
H2: LTL carriers rely less on the use of owner operators than do TL carriers.
In this research, we cannot observe uncertainty directly originating from the owner operator side (i.e., supplier opportunistic behavior): therefore, we cannot measure uncertainty occurring in the contractual relationship. However, we can measure uncertainty arising from the for-hire trucking carriers: revenue variations due to unpredictable market demand.
Sales volume uncertainty due to market demand volatility has been considered a good example of environmental uncertainty in the context of transaction cost economics (Shelanski and Klein 1995). Williamson's initial prediction that firms are more likely to integrate than to outsource when environmental uncertainty is high is supported by empirical evidence on the impact of uncertainty when measured by sales variation. For example, Walker and Weber (1984) used volume uncertainty in their model and predicted that volume uncertainty leads to making rather than buying based on the rationale that firms should be able to "coordinate swings internally" more efficiently than to manage variations with suppliers. MacMillan et al. (1986) also found that uncertainty increases integration (making) and reduces buying. In their study, sales instability was included as an uncertainty variable in the model to examine why firms adopt backward integration. Levy (1985) also included variance of unanticipated sales as one of the explanatory variables in the regression model for explaining firm vertical integration. He found that sales variance, considered a risk factor, leads firms toward more vertical integration. However, in contrast to the above studies, Harrigan (1986) used "change in sales growth" as a measurement of uncertainty and found that uncertainty reduces integration.
In line with Williamson's initial prediction, we argue that demand uncertainty requires trucking carriers to use company drivers due to three considerations. First, when contracts are subject to a fixed expiration date and other fixed terms, for-hire trucking carriers may not be able to gain flexibility in the presence of short-term demand changes. While using company drivers, for-hire trucking carriers may require drivers to work overtime in the presence of demand surges and reduce driver work hours or lay off drivers in the presence of low demand. Furthermore, it is observed that non-union carriers have more flexibility to move shifts around and often hire part-time employees who work on an as-needed basis to supplement various shifts (Fisher and Home 2004; Corsi 2005). Second, it appears to be a legitimate argument that the use of owner operators may add flexibility at first glance. However, a deep examination reveals that this may not necessarily be the case at all times. The business of for-hire trucking is highly correlated with the general economy and subject to seasonal variations as well. When the economy booms, every carrier wants to hire owner operators to complement the shortage of company drivers, resulting in a shortage of the market supply of drivers and equipment. Third, even with a significant use of owner operators, for-hire carriers cannot frequently adjust the usage by firing current owner operators or recruiting new owner operators, because searching for qualified owner operators and identifying safe and qualified trucking equipment is costly and time consuming and transaction costs associated with renegotiating and rewriting certain long-term supply contracts are prohibitive. In a turbulent environment, trucking carriers would be better off using company drivers instead of owner operators. Therefore, we propose the following hypothesis:
H3: The use of owner operators is negatively associated with the level of revenue uncertainty.
We draw on the structure-conduct-performance paradigm to explain one import linkage: how structure (measured by segment concentration) affects conduct (outsourcing decision in this case, measured by whether to use owner operators or hire company drivers). We believe that for-hire trucking carriers cannot make optimal decisions without considering structural constraints, especially segment competition. Spanos et al. (2004) argues that in industrial organization theory, industry concentration may be the most critical industry structural variable.
In general, for-hire trucking in the U.S. is a competitive business. For-hire trucking carriers compete for qualified drivers as well as for delivery business. The driver shortage has been exacerbated in the post 9-11 era due to increased standards for background checks, which has created significant problems for the motor carrier industry, especially for TL carriers, such as lost revenue, delayed services, or high driver turnover rates (Corsi 2005).
However, the U.S. for-hire trucking industry is not homogenous. In a more concentrated segment, a small number of large carriers can afford to retain a stable force of drivers due to higher rates shippers have to pay and increased firm profitability. When concentration is low, which means that the segment is more competitive, for-hire trucking carriers are not able to charge premium rates and may have to resort to the use of owner operators to cut back on operating costs. Using owner operators instead of company drivers, especially unionized company drivers, turns out to be a viable method for carriers to cut back on delivery costs. This is evidenced by the fact that the Teamsters Union considers low-wage owner operators as a threat to well-paid jobs for employee drivers and, therefore, seeks to constrain the use of owner operators by U.S. for-hire carriers (Peoples and Peteraf 1999). The ability to charge higher prices enables for-hire carriers to hire more company drivers and, therefore, mitigate the cost reduction pressure of using owner operators.
Trucking firms facing competition in the post-1980 period may have to price competitively due to rivalry from competing carriers and from other modes of transportation, such as rail (McMullen 2004). Meanwhile, evidence from the financial service industry suggests that industry competition impacts a firm's product differentiation strategy (Cohen and Mazzeo 2004). Here we argue that in a more concentrated segment, carriers have more financial resources to pursue a product differentiation strategy and, therefore, sustain premium prices. In contrast, in a more competitive segment, carriers are unlikely to charge high prices and have to compete on costs. In this latter scenario, low-wage owner operators become an important alternative. The success of a product differentiation strategy depends on whether carriers are able to provide superior services. In this study, a superior service is a multi-dimensional measure that includes shorter transit times, availability of real-time tracking, less damage, fewer accidents, professional conduct, brand names, etc. Due to the increased difficulty and costs in monitoring, coordinating, and motivating owner operators, carriers pursuing a product differentiation strategy may find it easier to achieve their objectives by using company drivers. Generally speaking, the use of owner operators works better for a low-cost strategy. Therefore, we propose the following hypothesis:
H4: The use of owner operators is negatively associated with the level of segment concentration.
Hypotheses 1, 2, and 3 are predicted by transaction cost economics, while Hypothesis 4 is predicted by the structure-conduct-performance paradigm. We summarize our hypotheses in Figure 1.
[FIGURE 1 OMITTED]
According to Figure 1, we expect that all transaction-cost-related variables (the use of specialized equipment, engagement in LTL services, and revenue uncertainty) reduce the use of owner operators, while the industry concentration level also reduces the use of owner operators. Additionally, we include a few firm-specific control variables, such as asset structure (leased equipment), operating profitability (operating ratio), equipment, operating productivity (miles per equipment), communications costs, and firm size. We believe that those variables are closely related to the use of owner operators who provide both equipment and labor. For example, asset structure reflects a carrier's equipment ownership structure and, to a great extent, reflects a carrier's operating strategy. Carriers who lease more equipment are less likely to maintain a large team of company drivers, and, therefore, will employ more owner operators in case of driver shortages. Carriers with high operating ratios may have a strong incentive to use owner operators as a possible means of cutting operating costs in many cases. The use of owner operators may increase communication costs due to the increased need for coordination. Finally, we believe that larger firms may have stronger financial resources and can afford to hire company drivers due to economies of scale, while smaller firms may have to rely on owner operators to cut back on operating costs. Therefore, firm size is an important control variable.
Sample and Data Collection
Our dataset is collected from the annual reports filed with the U.S. Department of Transportation by the largest U.S. interstate for-hire carriers. Interstate for-hire carriers generating more than $10 million revenue annually (Class I carriers) and generating above $3 million but less than $10 million annually (Class II carriers) are required to file annual financial and operations reports with the Department of Transportation under U.S. regulation 49 CFR 1420. Class III carriers generating annual revenue under $3 million are exempt from the filing requirements. Therefore, most carriers included in our dataset are relatively large players in the for-hire trucking industry.
Even though the original dataset has 2,137 entries, we are able to include 1,645 carriers with complete and accurate information on all variables in the final analysis as a consequence of reporting errors and missing values on certain variables. There were eighty-four observations excluded because some variables contained in the records were clearly reporting errors, such as negative values in revenue, assets, communications costs, and owner operator expenses, unusually small values in fleet size, etc. In terms of missing values, a common characteristic of archival data, we opted to exclude observations from the final sample if data were missing for any of the variables in the model to ensure the integrity of the original data and the quality of data analysis. Two variables were responsible for the missing data:
(1) Rev_Change (percentage change in revenue from 2002 to 2003) and (2) Average_Mile (average mile per piece of equipment). One hundred and ninety-seven firms did not file annual reports to the U.S. Department of Transportation in both 2002 and 2003; this is not surprising given the relatively high rates of for-hire carrier entry and exit. Three hundred and twenty-four firms did not report total miles and 117 firms did not report total fleet sizes. When all missing categories were combined, 408 records were excluded. Altogether, reporting errors and missing values result in an exclusion of 492 records from the final analysis.
The dependent variable in our analysis is the usage of owner operators (OWNER). In their financial reports, carriers report the expenses on equipment rentals with drivers. Since owner operators are generally defined as drivers who provide their own equipment, the percentage of this expense out of the total operating expenses reflects how owner operators are used relative to company drivers. In the financial reports filed to the Department of Transportation, the expense category of owner operators is clearly identified as equipment rental with drivers, and such expenses are reported on drivers' 1099 Form for the purpose of the U.S. IRS income tax returns.
Note that in our dataset, carriers identify themselves as one of two major segments: specialized freight carrier (SPE) or general freight carrier. Based on the portion of LTL revenue, general freight carriers are further classified as LTL carriers if the LTL business accounts for at least 50 percent of the total revenue, or TL otherwise. Since SPE, LTL, and TL are all binary dummies, in our regression model, only the SPE and LTL dummy variables are kept, with TL being the default missing dummy variable. Since specialized freight services require specialized equipment, it is a good measure of physical asset specificity. Meanwhile, since LTL business involves a high level of coordination and investments in hub-and-spoke network and technologies, LTL reflects higher asset specificity relative to TL services.
Volume uncertainty (REV_CHANGE) is measured as the absolute percentage change in revenue from the previous year of 2002. Its magnitude reflects the level of market uncertainty a carrier is facing. Segment concentration is based on the distribution of revenues among all the firms operating in each sub-segment. In this dataset, a total of eleven segments are reported: TL, LTL, and nine sub-segments under specialized freight, such as agricultural, building materials, bulk freight, household goods, heavy machinery, motor vehicle, refrigerated freight, tank truck, and other specialized. The market concentration ratio for each sub-segment is measured by the Herfindahl-Hirschman Index (HHI), which is calculated by summing up all the squared market shares in particular sub-segments. Generally speaking, a value under 1,000 may be considered more competitive, while a value over 2,000 may be considered highly concentrated. In fact, the U.S. Department of Justice adopts a more stringent rule in judging mergers and antitrust cases. In a case against Computer Associated International, Inc. (2001), the U.S. Department of Justice considered a market in which HHI is over 1,800 to be highly concentrated, a market with HHI under 1,000 to be more competitive, and a market in which HHI is between 1,000 and 1,800 to be moderately concentrated. In our study, HHI is normalized by dividing 10,000 into original HHI scores so that HHI stays between 0 and 1.
Besides the above dependent and independent variables, we include a few control variables in the regression model specified as follows: Operating ratio (OP_Ratio) is the ratio of total operating expenses and total operating revenues, reflecting operating profitability and basic financial conditions. Leased equipment (Leased_Equip) is the ratio of number of leased equipment over total equipment, reflecting a firm's asset ownership structure. Communication costs (COMM) is measured as actual communications costs as a percentage of total operating costs. In our dataset, communication costs include expenses spent on long distance, fax, telephone equipment and services, etc. Average mile operated per piece of equipment (Average_Mile) is calculated by the total miles over the total amount of equipment, an indicator of operating efficiency. It is expressed in thousand miles per piece of equipment. Total revenue in hundred million dollars (Mrevenue) is used as a size measure.
Based on the reasoning laid out above, the outsourcing decision is an outcome of analysis of internal resources (as described by transaction cost economics) and external structure (as described by the structure-conduct-performance paradigm), and other firm-specific factors. We assume that the use of owner operators is a linear function of the above factors, which can be expressed as follows:
Use of owner operators = f(transaction cost variables, structural variable, firm controls). With a normality assumption, we start with an OLS model to estimate the use of owner operators. The full regression model is specified as follows:
Owner = [[beta].sub.0] + [[beta].sub.1] x SPE + [[beta].sub.2] x LTL + [[beta].sub.3] x Rev_Change + [[beta].sub.4] x HHI + [[beta].sub.5] x OP_Ratio + [[beta].sub.6] x COMM + [[beta].sub.7] x Leased_Equip + [[beta].sub.8] x Average_Mile + [[beta].sub.9] x MRevenue + [epsilon]
Descriptive Statistics and Correlation Table
Table 1 shows that the extent to which owner operators are used varies significantly across for-hire carriers. Some carriers did not use owner operators at all, while others may be all owner operator carriers. Revenue change (REV_Change) varies. Some carriers had a very stable revenue stream (a nearly 0% percent change), while others may double revenue (a 97 percent increase). Segment concentration ratio (HHI), if not divided by 10,000, ranges from 250 to 3,348. Therefore, it appears that some segments are highly concentrated while others are more competitive. The operating ratio (OP_Ratio) ranges from 73 percent to 210 percent. A low operating ratio indicates efficient operations and stronger financials, while a high operating ratio suggests otherwise. The maximum value of 2.10 indicates that this carrier's expense more than doubled its total operating revenue. Firm asset structure varies significantly. Some firms own all equipment, while others rent all equipment. Overall, the average ownership structure is 30 percent rental vs. 70 percent ownership. Communication costs are a very tiny portion of the total operating expenses (less than 1 percent on average) and variation is small across carriers. On average, each piece of equipment travels nearly 39,000 miles with dramatic variations across carriers. Firm size in terms of total operating revenue varies significantly. While the average revenue is approximately $52 million, smaller carriers may have revenues less than $ 3 million and giant carriers generate revenues as large as $15.9 billion in the sample.
Table 2 shows that all independent variables are moderately correlated with each other.
Table 3 provides the results of the OLS estimation. The coefficient for SPE is -0.0414 with (p<.05), indicating that the use of specialized equipment reduces the use of owner operators and that H1 is supported. The coefficient for LTL is -0.0805 with (p< .01), indicating that the engagement in LTL services reduces the use of owner operators and that H2 is supported. The coefficient for Rev_Change is far from significant, indicating that H3 is not supported. SPE, LTL, and REV_Change combined suggest that transaction cost variables make a relatively strong contribution to account for a carrier's outsourcing decision in the presence of other variables. The coefficient for HHI (segment concentration) is -0.1249 with (p<.01), demonstrating that segment concentration reduces the use of owner operators. In other words, the more concentrated the market is, the less pressure a carrier is likely under to use owner operators, ceteris paribus. This result provides a strong support for H4.
This OLS model has a significant F-value of 39.88 and an adjusted R-squared of 17.55 percent, suggesting a reasonable explanatory power given the nature of a cross-sectional dataset being used. This overall explanatory power, combined with the significant coefficients for most of the independent variables, indicates that transaction cost economics with the help of the structure-conduct-performance paradigm can have a strong explanatory power for a carrier's outsourcing decision.
One of the critical assumptions for the classical OLS model is that variables including the dependent variable are normally distributed. Violation of this normality assumption is likely to result in biased estimations. Note that in our dataset, nearly one-third of the values of the dependent variable are zeros, resulting in a left-censoring of the sample, while the use of owner operators may appear to be normal for values beyond zero and less than 100 percent. When data are censored, possible observations beyond the cutting point may be truncated and only part of the distribution is used in the analysis. To make up for a full probability distribution, the censored regression model (e.g., a Tobit model) has been introduced to address this issue by correcting the distribution (Greene 2003, 762-766). The adjusted regression results are reported in Table 4.
Comparing the results of the OLS model (Table 3) with those of the Tobit model (Table 4), we note that these two models have a high level of agreement. All the signs of independent variables are in the same directions. More importantly, when corrections are made for the left-censored values using the Tobit model, the significance levels of all independent variables (SPE, LTL, and HHI) have been enhanced, providing strong supportive evidence for our hypotheses H1, H2, and H4. As Table 4 shows, the coefficient for SPE is -0.0741 with (p<.01), -0.1515 with (p<.01) for LTL, and -0.1996 for HHI with (p<.05). The magnitudes of those coefficients are much larger than what were found in the OLS model. Most control variables, such as operating ratio, leased equipment, communications, and average miles driven per piece of equipment, show similar impacts as in the OLS model.
DiscussioN OF RESULTS
Note that the coefficient for LTL services has a larger magnitude than that for specialized equipment, and the significance level of the LTL coefficient (only t-stats are reported in Tables 3 and 4) is much higher than that of specialized equipment across both models. Taking results from the Tobit model, the coefficient for specialized equipment (SPE) is -.0741 with (t =-2.86), suggesting that being a specialized carrier, holding everything else constant, reduces the use of owner operators by more than 7 percentage points: while the coefficient for LTL services (LTL) is -.1515 with (t = -5.70), suggesting that being a LTL carrier, holding everything else constant, deceases the use of owner operators by more than 15 percentage points. This direct comparison may suggest that engagement in the LTL services requires more vertical integration than specialized equipment does. Such a comparison is reasonable in that the risk associated with specialized equipment is much lower than that associated with the LTL services. Even though specialized equipment is difficult to be redeployed due to physical asset specificity, it is still mobile and can be resold as well. With this salvage value in mind, we believe that the costs for specialized equipment investment are much lower than investments in the LTL services. As analyzed earlier, LTL services require high upfront fixed investment in the establishment of the hub-and-spoke network, a high level of coordination among drivers, between management and drivers, and between drivers and customers. Also note that the coefficient for HHI is -.1996 with (p<.05); since HHI is normalized by a base of 10,000, we would expect a 1,000 increase in the segment concentration ratio would reduce the use of owner operators by about 2 percentage points.
The study did not find a significant impact of revenue change on the use of owner operators across both the OLS and Tobit models, as the transaction cost economics theory expects. A possible explanation is that we operationalized the extent of uncertainty facing for-hire carriers by revenue percentage change over a prior year (REV_Change), which may not be able to capture the characteristics of environmental uncertainty. Unusual events, such as mergers and divestures, have a direct impact on revenues and may distort the actual picture of uncertainty.
In summary, the results indicate that the decision to use company drivers or owner operators is more likely to be driven by transaction cost consideration and market competition. Except the existence of union drivers in certain segments, the institutional environment facing U.S. for-hire carriers is much less of a constraint on their employment decisions. The U.S. trucking industry may differ from its European counterpart, as Arrunada et al. (2004) pointed out. In fact, U.S. for-hire trucking companies have been faced with less strict labor and tax regulation since deregulation and company driver turnover has been prevalent. The results further suggest that U.S. LTL carriers strongly favor company drivers over owner operators. In contrast, the use of owner operators is very popular in the European LTL services, especially in Spain. Fernandez et al. (2000) explained that "quasi-integration" of Spanish owner operators provides a more efficient organization form for LTL carriers than vertical integration.
Transaction cost economics makes a strong statement that asset specificity increases market transaction costs and prefers in-house production or higher level of vertical integration. Based on for-hire trucking data, this study finds that the use of specialized equipment and the engagement in LTL business reduce the use of owner operators compared to general freight and non-LTL operations, ceteris paribus. Such results provide a strong support for Williamson's core notion of asset specificity. In our current study, higher transactions costs associated with frequent search for qualified owner operators push carriers to seek higher level of integration by using more company drivers. Meanwhile, the structure-conduct-performance paradigm posits that firm decisions are constrained by the industry structure, most prominently the concentration level. This current study shows that concentration levels reduce the use of owner operators, suggesting that higher concentration may lead to a higher level of vertical integration.
Our contribution to the outsourcing literature is that we complement transaction cost economics with an industry structural constraint and test this integrated framework using a most recently available and large-scale cross-sectional dataset. We found that both transaction-cost-related factors and segment concentration have significant impacts on for-hire carriers' decisions to use owner operators. The structural variable turns out to be a valuable complement to the classical transaction cost variables in accounting for carrier outsourcing decisions.
We also provide managerial implications in general and to the for-hire trucking in particular that outsourcing is not a bandwagon decision and management should consider both relationship-specific transaction cost parameters and industry structural characteristics. The U.S. trucking industry practice suggests that LTL carriers tend to use more of company drivers and less of owner operators to ensure the smooth operations mandated by the complex hub-and-spoke network. Also, when carriers are engaged in transporting specialized freight, they tend to reduce the use of owner operators.
We did not find support for the uncertainty argument of transaction cost economics. As we discussed earlier, this may be due to the fact that uncertainty is measured by revenue percentage change over the prior year. Standard deviation of revenues over a longer period or other types of measures may be preferred in future research. We are not able to operationalize transaction frequency, one of the three constructs of the classic transaction cost economics, due to data constraints, a problem faced by the majority of the existing transaction cost economics literature. Also note that certain segments of large for-hire carriers may be unionized, but unionization information is not available in this dataset. Further research may consider controlling for such conditions in analysis when data become available. We admit that there may exist a potential endogeneity issue in the relationship between subcontracting decision and segment concentration. On one hand, segment concentration constrains a carrier's decision to subcontract in the current period since carriers are concerned about their service quality under competitive pressure; on the other hand, subcontracting may affect carrier service quality, which subsequently affects a carrier's market share in the near future and finally impacts segment concentration. Future research may be able to address this issue using data over a longer period. Last, but not least, we are aware that we are constrained by the current data to include only a few firm-specific variables as control variables in the econometric model. Other factors, such as national and regional macroeconomics, characteristics about the owner operators, firm history, management style, and operating strategies, may have a significant impact on the use of owner operators but are not captured in the current model. Therefore, our model may be subject to underspecification and potential estimation biases.
Arrunada, B., M. Gonzalez-Diaz, and A. Fernandez (2004), "Determinants of Organizational Form: Transaction Costs and Institutions in the European Trucking Industry," Industrial and Corporate Change, 13(6): pp. 867-882.
Baker, G. P. and T. N. Hubbard (2004), "Contractibility and Asset Ownership: On-board Computers and Governance in U.S. Trucking," The Quarterly Journal of Economics, November, pp. 1443-79.
Coase, R. H.(1937 and reprinted in 1991), "The Nature of the Firm," in Williamson, O.E., and S. G. Winter (eds.), The Nature of the Firm. Origins. Evolution, and Development. Oxford University Press, New York.
Cohen, A. and M. J. Mazzeo (2004), "Competition, Product Differentiation and Quality Provision: An Empirical Equilibrium Analysis of Bank Branching Decisions," Finance and Economics Discussion Series, Federal Reserve Board, Washington, DC. http:// www.federalreserve.gov/pubs/FEDS/2004/200446/200446 pap.pdf, accessed April 7, 2008.
Corsi, T. M. (2005), "Motor Carrier Industry Profile (2004-2005)," Robert. H. Smith School of Business, University of Maryland, pp. 1-34. http://ai.volpe.dot.gov/ CarrierResearchResults/PDFs/Motor_Carrier_Industry_ Profile_Update_2004-2005-final.pdf, accessed November 6, 2007.
Fernandez, A., B. Arrunada, and M. Gonzalez-Diaz (2000). "Quasi-Integration in Less-than-Truckload Trucking," in C. Menard (ed.), Institutions. Contracts. Organizations: Perspectives from New Institutional Economics, Edward Elgar, Cheltenham, pp.293-312.
Fisher, W. H. and B. Home (2004), "Less Than Truckload: Key Secular Trends Driving Regional Carriers," Raymond James & Associates, Inc., October 13, 2004.
Greene, W. H. (2003), Econometric Analysis, 5th Edition, Pearson Education Inc.
Guido, D. W. (2005), "Driver Churn Sets Record," Transport Topics, April 4. 2005, p. 1 & p.26.
Harrigan, K. R. (1986), "Matching Vertical Integration Strategies to Competitive Conditions," Strategic Management Journal, 7(6): pp. 535-555.
He, D. and J.A. Nickerson (2006), "Why Do Firms Make and Buy? Efficiency, Appropriability, and Competition in the Trucking Industry," Strategic Organization, 4(1): pp. 43-69.
Klein, B., R. Crawford, and A. Alchian (1978), "Vertical Integration, Appropriable Rents, and the Competitive Contracting Process," Journal of Law and Economics, 21 (October): pp.297-326.
Koopmans, T. (1957), Three Essays on the State of Economic Science, McGraw-Hill, New York.
Levy, D. T. (1985), "The Transactions Cost Approach to Vertical Integration: An Empirical Examination," Review of Economics and Statistics, 67, pp. 438-445
MacMillan, I. C., D. C. Hambrick, and J. M. Pennings (1986). "Uncertainty Reduction and the Treat of Supplier Retaliation: Two Views of the Backward Integration Decision," Organization Studies, 7(3): pp. 263-278.
McMullen, B. Start (2004), The Evolution and Current Market Structure of the U.S. Motor Carrier Industry, http://oregonstate.edu/-mcmulles/new465/trucking.htm, accessed April 5, 2008.
Nickerson, J. A. and B.S. Silverman (2003), "Why Aren't All Truck Drivers Owner-Operators? Asset Ownership and the Employment Relation in Interstate For-Hire Trucking," Journal of Economics & Management Strategy, 12(1): pp. 91-118.
Owner Operator Independent Driver's Association (2008), OOIDA By-laws, http://www.ooida.com/ ooidamembers/By_Laws/by_laws_120606.pdf, accessed April 5, 2008.
Peoples, J. and W. K. Talley (2007), "Owner-Operators Truck Driver Earnings and Employment: Port Cities and Deregulation," Old Dominion University, http:// bpa.odu.edu/port/research/research.shtml, accessed May 2, 2007.
Peoples, J. and M. Peteraf (1999), "The Effects of Regulation Reform on Company Drivers and Owner Operators in the For-hire and Private Sectors," Transportation Journal, Spring, Vol. 38, No. 3, pp. 5-17.
Shelanski, H. A. and P. G. Klein (1995), "Empirical Research in Transaction Cost Economics: A Review and Assessment," Journal of Law, Economics and Organization, 11(2): pp. 335-361.
Spanos, Y. E., G. Zaralis, and S. Lioukas (2004), "Strategy and Industry Effects on Profitability: Evidence from Greece," Strategic Management Journal, 25, pp. 139-165.
U.S. Census Bureau (2002), North American Industry Classification System, Trucking, http://www.census.gov/ epcd/naics02/def/NDEF484.HTM, accessed April 13, 2008.
U.S. Department of Justice (2001), Antitrust Division, USA v. Computers Associates International, Inc., http:// www.usdoj.gov/atr/cases/f0300/0306.htm, accessed April 10, 2008.
Waldman, D.E. and E. J. Jensen (2006), Industrial Organization, 3rd Edition, Pearson Education, Inc., Boston.
Walker, G. and D. Weber (1984), "A Transaction Cost Approach to Make-or Buy Decisions," Administrative Service Quarterly, 29(3): pp. 373-391.
Williamson, O.E. (2005), "The Economics of Governance," The American Economic Review, AEA Papers and Proceedings, 95(2): pp.1-18. Williamson, O. E. (1985), The Economic Institution of Capitalism, The Free Press, New York.
Williamson, O. E. (1975), Markets and Hierarchies, The Free Press, New York.
Mr. Han is a doctoral student, Department of Logistics, Business and Public Policy, Robert H. Smith School of Business, University of Maryland, College Park, Maryland 20742; Mr. Corsi, EM-AST&L, is Michelle Smith Professor of Logistics and Co-Director, Supply Chain Management Center, Robert H. Smith School of Business, University of Maryland; Mr. Grimm, EMAST&L, is Dean's Professor of Supply Chain & Strategy, Robert H. Smith School of Business, University of Maryland.
Table 1. Descriptive Statistics Standard Variable Mean Minimum Maximum Deviation Owner 0.1878 0.0000 0.9831 0.2427 SPE 0.4274 0.0000 1.0000 0.4877 LTL 0.0802 0.0000 1.0000 0.2735 REV_Change 0.1020 0.0001 0.9865 0.1075 HHI 0.1036 0.0250 0.3348 0.0517 OP_Ratio 0.9859 0.7347 2.1017 0.0759 COMM 0.0101 0.0000 0.1209 0.0075 Leased_Equip 0.2694 0.0000 1.0000 0.3506 Average_Mile 38.9233 0.4751 401.7000 33.3444 Mrevenue 0.5159 0.0219 159.5009 4.4814 Table 2. Correlations of Independent Variables 1 2 3 1 SPE 1.00 2 LTL -0.24 *** 1.00 3 REV Change 0.01 -0.02 1.00 4 HHI -0.29* * -0.12 *** 0.02 5 OP Ratio 0.03 -0.04 0.05 ** 6 COMM 0.02 0.17 *** -0.01 7 Leased Equip -0.08 *** -0.05 ** 0.02 8 Average Mile -0.04 -0.12 *** 0.08 *** 9 Mrevenue 0.02 0.05 * -0.02 4 5 6 1 SPE 2 LTL 3 REV Change 4 HHI 1.00 5 OP Ratio -0.02 1.00 6 COMM 0.03 0.07 *** 1.00 7 Leased Equip -0.05 * 0.04 * -0.12 *** 8 Average Mile -0.07 *** -0.05 * -0.14 ** 9 Mrevenue 0.03 -0.04 * 0.01 7 8 9 1 SPE 2 LTL 3 REV Change 4 HHI 5 OP Ratio 6 COMM 7 Leased Equip 1.00 8 Average Mile 0.06 *** 1.00 9 Mrevenue -0.03 -0.01 1.00 (Note: pearson correlation coefficients, * denotes p<.1, ** p<.05 and *** p<.01) Table 3. Regression Results of the Baseline OLS Model Independent variables meaning SPE specialized equipment dummy LTL LTL services dummy REV_Change absolute value of revenue percentage change HHI Herfindal and Hirschman Index divided by 10,000 OP_Ratio operating expenses over total revenue COMM communications expenses over total expenses Leased_Equp number of leased equipment over total equipment Average_Mile thousand miles operated per piece of equipment Mrevenue revenue in 100 million dollars Independent standard variables coefficient error t-statistics SPE -0.0414 0.0196 -2.12 ** LTL -0.0805 0.0191 -4.22 *** REV_Change 0.0377 0.0508 .74 HHI -0.1249 0.0753 -1.66 * OP_Ratio -0.0209 0.0687 -.30 COMM -3.4276 0.7505 -4.57 *** Leased_Equp 0.1868 0.0157 11.87 *** Average_Mile 0.0013 0.0002 7.89 *** Mrevenue -0.0015 0.0012 -1.19 Number of observations 1645 F-value 39.88 p-value <0.0001 R-squared 0.1800 Adjusted R-squared 0.1755 (two-tailed tests, * denotes p<.1, ** p<.05 and *** p<.01) Table 4. Regression Results of the Tobit Model Dependent variable: use of owner operators Independent variables meaning SPE specialized equipment duty LTL LTL services duty REV_Change absolute value of revenue percentage change HHI Herfindal and Hirschman Index divided by 10,000 OP_Ratio operating expenses over total revenue COMM communications expenses over total expenses Leased_Equp number of leased equipment over total equipment Average_Mile thousand miles operated per piece of equipment Mrevenue revenue in 100 million dollars Dependent variable: Independent standard variables coefficient error t-statistics SPE -0.0741 0.0259 -2.86 *** LTL -0.1515 0.0266 -5.70 *** REV_Change -0.0078 0.0682 -0.11 HHI -0.1996 0.1008 -1.98 OP_Ratio -0.0531 0.0932 -0.57 COMM -3.7674 0.9825 -3.83 * Leased_Equp 0.2247 0.0205 10.95 *** Average_Mile 0.0016 0.0002 7.38 *** Mrevenue -0.0081 0.0042 -1.93 * Number of observations 1645 Number of left-censored values 456 Log Likelihood -513.30 (two-tailed tests. * denotes p<.1, *** p<.05 and *** p<.01)