Fuel Economy and Safety

Transcription

1 Fuel Economy and Safety A Reexamination under the U.S. Footprint-Based Fuel Economy Standards Jiaxi Wang University of California, Irvine

2 Abstract The purpose of this study is to reexamine the tradeoff between fuel economy and automobile safety under the recently amended U.S. Corporate Average Fuel Economy (CAFE) standards. In formulating the model, the framework consists of two submodels: a fuel economy submodel and a vehicle safety submodel. The empirical results indicate that (1) footprint and weight are likely to have different roles in in determining fuel economy and vehicle safety; (2) for passenger cars, increases in footprint do not decrease vehicle fuel economy, however, decrease the likelihood of more severe occupant injuries; (3) for light trucks, a higher weight decreases fuel economy slightly, yet increases safety for its occupants markedly.

3 Background Responding to criticisms that the old CAFE encourages the production of smaller vehicles, which unfavorably impacts domestic automakers compared to foreign automakers and may also increase traffic safety risks, starting in 2011, the CAFE standards are newly expressed as mathematical functions depending on vehicle footprint. The mathematical function of new CAFE standards provides a separate fuel economy target for each vehicle footprint, generally with more stringent targets for smaller vehicles and less stringent targets for larger vehicles. Some critics pointed out that the new standards might have had the unintended consequence of pushing manufacturers to make everlarger vehicles to avoid strict economy standards.

4 Research Questions How do vehicle footprint and weight affect fuel economy, respectively? Do vehicle footprint and weight are a proxy to each other in determining vehicle safety? Whether increased safety in modern vehicles has been an obstacle for automakers to fulfil the new CAFE standards?

5 Data Vehicle Crash Data The primary basis for the data set in the vehicle safety study of this paper is the General Estimates System (GES) data produced by the National Automotive Sampling System (NASS). GES data come from a nationally representative sample of police reported motor vehicle crashes of all types, from minor to fatal. The GES data set consists of three separate files: vehicle, crash, and person. These were combined on the basis of the person-level data set s sample of drivers. To study vehicles that were produced under the newly amended footprint-based CAFE standards, the data set was further limited to vehicles after the 2011 Model Year.

6 Data Vehicle Attributes Data Since the GES dataset includes vehicle information such model year, make, model, and body type, specific vehicle models can be identified for all vehicles. For each single-vehicle crash for which occupant injury severity was available, information on the involved vehicle s attributes including but not limited to footprint, curb weight, horsepower, torque, transmission, and Manufacturer Suggested Retail Price (MSRP) were obtained from For accuracy, vehicle fuel economy data were obtained from the official U.S. government source for fuel economy information.

7 Summary Statistics for Vehicle Attributes at Model Level Variable Mean Std. Dev. Min Max Passenger Cars Fuel Economy Footprint Curb Weight Height Length MSRP Horsepower Torque Automatic Hybrid Sample Size 420 Light Trucks Fuel Economy Footprint Curb Weight Height Length MSRP Horsepower Torque Automatic Sample Size 185

8 Fuel Economy Submodel Empirical Strategy The fuel economy of a vehicle i produced at time t (mpg it ) was modeled as a function at a given level of footprint (fp it ), curb weight (w it ), and other attributes related to fuel economy (X it ). T t is a year dummy variable used to control for possible fuel-efficiency technologies progress in automobile industry. ε it is a mean zero error capturing additional characteristics of the vehicle that are assumed to be uncorrelated with the other right-hand-side variables. Under Cobb-Douglas assumptions, fuel economy is modeled as (1) ln mpg it = β 1 ln fp it + X it B + T t + ε it (2) ln mpg it = β 2 ln w it + X it B + T t + ε it (3) ln mpg it = β 1 ln fp it + β 2 ln w it + X it B + T t + ε it

9 Fuel Economy Submodel Passenger Cars A 10% increase in footprint is associated with a 1.71% increase, rather than a decrease, in fuel economy or, equivalently, as footprint increases by 100 ft 2, the average fuel economy for passenger cars increases by 0.06 MPG. A 10% increase in curb weight is associated with a 5.27% decrease in fuel economy, translating that a car that is 100-pound heavier is 0.39 MPG less fuel efficient than is its lighter counterpart. Light Trucks Vehicle footprint does not have any significant effect on fuel economy for light trucks. A 10%increase in curb weight is associated with a 5.06% decrease in fuel economy; in other words, a light truck that is about 100 pounds heavier is only 0.23 MPG less fuel efficient when compared with its lighter counterpart.

10 Fuel Economy Submodel Trade-Off Estimates for Passenger Cars Model 1 Model 2 Model 3 Model 4 Model 5 Model 6 ln(footprint) *** (0.081) ** (0.078) 0.207** (0.104) 0.171* (0.096) ln(weight) *** (0.059) *** (0.057) *** (0.080) *** (0.077) ln(hp) (0.089) (0.094) (0.090) (0.094) (0.091) (0.095) ln(torque) ** (0.084) ** (0.087) (0.087) (0.090) (0.089) (0.091) ln(msrp) (0.031) (0.030) (0.034) (0.033) (0.036) (0.035) Automatic (0.010) (0.010) (0.009) 0.015* (0.008) (0.009) (0.009) Hybrid 0.413*** (0.033) 0413*** (0.033) 0.443*** (0.040) 0.441*** (0.039) 0.441*** (0.039) 0.439*** (0.039) Year Fixed Effects No Yes No Yes No Yes Observations Adjusted R Note: Standard errors are reported in parenthesis. *** Significant at the 1 percent level. ** Significant at the 5 percent level. * Significant at the 10 percent level.

11 Fuel Economy Submodel Trade-Off Estimates for Light Trucks Model 1 Model 2 Model 3 Model 4 Model 5 Model 6 ln(footprint) -0269*** (0.076) *** (0.075) (0.067) (0.968) ln(weight) *** (0.084) *** (0.083) *** (0.093) *** (0.092) ln(hp) 0.206* (0.108) 0.187* (0.109) 0.243** (0.105) 0.225** (0.105) 0.237** (0.106) 0.219** (0.106) ln(torque) *** (0.099) *** (0.099) *** (0.098) *** (0.097) *** (0.010) *** (0.099) ln(msrp) *** 0.120*** 0.108*** 0.111*** (0.036) (0.036) (0.035) (0.035) (0.035) (0.035) Automatic 0.070** (0.031) 0.069** (0.032) 0.074*** (0.027) 0.074*** (0.027) 0.079*** (0.028) 0.079*** (0.029) Year Fixed Effects No Yes No Yes No Yes Observations R Note: Standard errors are reported in parenthesis. *** Significant at the 1 percent level. ** Significant at the 5 percent level. * Significant at the 10 percent level.

12 Fuel Economy Submodel Technological Progress Estimates All the technological progress coefficients are positive and most of them are increasing over time. Some of them are statistically significant while some of them are not. In general, we can conclude that there has been a technological progress during the sample period. As a result, we can expect that technological progress would improve automobile fuel efficiency without decreasing much to footprint or weight.

13 Fuel Economy Submodel Technological Progress Estimates for Passenger Cars and Light Trucks Passenger Cars Light Trucks Year Model 2 Model 4 Model 6 Model 2 Model 4 Model * (0.012) (0.012) (0.012) (0.015) (0.014) (0.014) *** (0.011) 0.050*** (0.011) 0.049*** (0.011) 0.035** (0.016) 0.031** (0.014) 0.031** (0.014) ** 0.056*** 0.056*** (0.026) (0.022) (0.021) (0.021) (0.017) (0.017)

14 Vehicle Safety Submodel Empirical Strategy Ordered probit regression models were estimated for occupant injury status in single-vehicle crashes for passenger cars and lightduty trucks, respectively. The latent and continuous measure of injury severity faced by occupants of vehicle i in a crash occurred at year t (Y it ) was modeled as a function of footprint (fp it ), curb weight (w it ), and three vectors of controlling variables describing vehicle attributes (V it ), crash environments (C it ), and driver characteristics (D it ).

15 Vehicle Safety Submodel Empirical Strategy The following specification was used Y it = α 1 fp it + α 2 w it + α V V nt + α C C it + α D D it + T t + ε it. The observed and coded discrete injury severity variable, Y it, is determined from the model as follows Y it = {0 if Y it μ 1 No Inury, 1 if μ 1 < Y it μ 2 Possible Injury, 2 if μ 2 < Y it μ 3 Nonincapacitating Evident Injury, 3 if μ 3 < Y it μ 4 Incapacitating Injury, 4 if μ 4 < Y it (Fatal Injury)} where the μ s are the thresholds to be estimated.

16 Vehicle Safety Submodel Passenger Cars For a 100 ft 2 increase in footprint, the odds of fatal injury versus the combined incapacitating injury, non-incapacitating evident injury, possible injury, and no injury categories are time lower. Curb weight does not have any positive effect on vehicle safety for passenger cars, which is consistent with findings from recent U.S. and internationals studies on modern cars. Light Trucks Footprint does not have any significantly positive effect on vehicle safety for light trucks. For a 100-pound increase in curb weight, the odds of fatal injury versus the combined incapacitating injury, non-incapacitating evident injury, possible injury, and no injury categories are time lower.

17 Vehicle Safety Submodel Results for Passenger Cars: Odds Ratio Model 1 Model 2 Model 3 Model 4 Model 5 Model 6 Footprint (per 100 ft 2 ) 0.969** (0.015) 0.969*** (0.015) 0.956** (0.020) 0.955** (0.020) Curb Weight (per 100 lbs) (0.021) (0.021) (0.031) (0.031) Static Stability Factor * * (30.033) (24.061) (15.794) (55.188) (47.127) Horsepower 1.007** (0.003) 1.007** (0.003) 1.008** (0.003) 1.007** (0.003) 1.007** (0.003) 1.007** (0.003) Torque 0.994* (0.003) 0.995* (0.004) 0.993* (0.004) 0.994* (0.004) 0.993* (0.004) 0.994* (0.004) MSRP (0.000) (0.000) (0.000) (0.000) (0.000) (0.000) Speed 2.607*** (0.462) 2.674*** (0.475) 2.631*** (0.467) 2.698*** (0.482) 2.621*** (0.465) 2.697*** (0.481) Weekday (0.105) (0.105) (0.105) (0.105) (0.105) (0.106) Afternoon 0.676*** (0.088) 0.681*** (0.089) 0.677*** (0.088) 0.682*** (0.089) 0.677*** (0.089) 0.682*** (0.090) Night (0.153) (0.154) (0.153) (0.154) (0.154) (0.155) Snow (0.407) (0.404) (0.406) (0.403) (0.407) (0.403) Rain (0.218) (0.218) (0.221) (0.221) (0.218) (0.218) Fog (0.602) (0.582) (0.601) (0.583) (0.613) (0.594) Low (0.194) (0.196) (0.193) (0.195) (0.194) (0.196) Middle (0.215) (0.215) (0.213) (0.213) (0.214) (0.215) Male (0.104) (0.103) (0.104) (0.103) (0.104) (0.103) Old 1.831*** (0.272) 1.829*** (0.273) 1.823*** (0.271) 1.821*** (0.272) 1.826*** (0.271) 1.824*** (0.272) Negligence 3.343*** (0.537) 3.364*** (0.543) 3.330*** (0.537) 3.352*** (0.543) 3.361*** (0.540) 3.388*** (0.547) μ μ μ μ Year Fixed Effects No Yes No Yes No Yes Observations LRI

18

19 Conclusion Passenger Cars The empirical results suggest that, for passenger cars, increases in footprint do not adversely affect fuel economy but decrease the likelihood of more sever occupant injuries in single-vehicle crashes; while the weight variable has a negative impact on fuel economy but no significant effect in occupant protection. Together, one can expect that increasing footprint, while using lighter materials in automobile manufacturing to reduce weight, makes it possible for society to increase both fuel economy and vehicle safety for passenger cars.

20 Conclusion Light Trucks The fuel economy trade-off estimates for light trucks suggest that only weight but not footprint has an small adverse effect on fuel economy. The ordered logistic regression results show that footprint does not have any significantly positive effect on vehicle safety, while curb weight help light trucks to protect their own occupants. Together, these empirical results suggest that active and passive measures developed in recent years seem have largely improved light trucks safety performance without adding much to their curb weight.

21 Thank you.