Sales of Mid-Size Sport Utility

¶ … sales of Mid-size Sport Utility Vehicles (SUVs) will be forecast using median U.S. gasoline price as the independent variable. The emerging strength of crossover SUVs will also be included in the analysis. Crossover SUVs are by definition unibody in construction, therefore have a better mileage rating than their larger competitors in the mid-size SUV class, which is primarily comprised of SUVs build on truck frames, resulting in the term Body-on-Frame. These larger frame-based SUVs get lower mileage as a result of their construction yet can two in excess of 5,000 lbs., which is a major reason why consumers still purchase them despite the low mileage they attain.

An assessment of each of the forecasting methods is provided in this paper, and R.L. Polk's Insight database has been used for tabulating vehicle registrations of the two groups of SUVs included in this forecasting analysis. These two groups are comprised of the vehicles that fit the descriptions of Body-on-Frame and unibody or cross-over SUV, and are defined below.

Body on Frame SUVs (more towing power, less fuel efficient) included in the Analysis

CHEVROLET - TRAIL BLAZER

CHEVROLET - TRAILBLAZER EXT

CHEVROLET - TRAILBLAZER SS

FORD - EXPLORER

GMC - ENVOY XL DENALI

KIA - SORENTO

NISSAN - PATHFINDER

TOYOTA - 4 RUNNER

Unibody SUVs (less towing power, more fuel efficient) included in the Analysis

CHEVROLET - EQUINOX

FORD - FREESTYLE

HONDA - PILOT

HYUNDAI - SANTA FE

NISSAN - MURANO

SUBARU - B9 TRIBECA

TOYOTA - HIGHLANDER

Methodology

Phase 1

R.L. Polk's Insight database was first used to gather the last twelve months of new vehicle sales and registrations for the two groups of SUVs, the median U.S. gas price using the Monthly Energy Review (2006). The hypotheses of this forecast was to determine if demand for the larger and less fuel efficient Body-on-Frame SUVs would be adversely impacted by rising gas prices in 2006, and what the resulting forecast in these larger vehicles would be. The secondary hypothesis is to determine the impact of gas price increases on the demand for fuel efficient SUVs, which are the unibody models defined in this analysis, and widely regarded in the industry as the next generation of SUVs.

Phase 2

The fully test the forecast hypothesis of unibody SUVs being more positively influenced by U.S. gasoline prices, the unibody group of vehicles shown earlier in this paper are correlated to U.S. gas prices as reported by the Monthly Energy Review (2006). Specifically looking for any positive negative correlation as the basis of forecasting strategies is the end result of Phase 2, where correlation results are used as the foundation of further statistical analyses and forecasting using Microsoft Excel's advanced statistical functions and Statistical Package for the Social Sciences (SPSS) Student Edition Version 13.

Tracking Time Series Behavior of Variables

As a first step to Phase 1 of this forecasting project, the historical behavior of all three variables are analyzed, and presented in the chart, Influence of Gas prices on Cross-Over (Unibody) and Body-on-Frame SUVs, which is based on an analysis of R.L. registration data for the last fourteen months.

What's immediately apparent from the graphic Influence of Gas prices on Cross-Over (Unibody) and Body-on-Frame SUVs is that there is intuitively speaking only slight increases in Unibody SUV demand given the rise in gas prices from January, 2006 through July, 2006. A 3-month forward moving average is also applied to gas prices to further smooth the trending effects of this independent variable. The result is also shown in Influence of Gas prices on Cross-Over (Unibody) and Body-on-Frame SUVs, which highlights what appears to be a slight cyclicality of gas prices that were the basis of the U.S.' And world's highest gas prices in April, 2006. The 3-month moving average of gas prices accentuates cyclicality and shows graphically that as an independent variable, there is some influence on Body-on-Frame and Crossover (Unibody) SUV demand, yet intuitively it doesn't appear to auto-correlate strongly with each other. Moving averages in generate are excellent forecasting tools for defining a median point for a forecast range, keeping in mind that a range of values is a more reasonable assumption from a moving average forecast than a precise, pint-point prediction value.

Ascertaining the strength of the relationship between gas prices and the demand for both Body-on-Frame and Unibody or Cross-over SUVs needs to be the foundation of trying to define with greater accuracy future sales of both classes of vehicles. In defining this foundation, correlation analysis is completed for both classes of vehicles as the dependent variable, and gas prices being the independent variable. The results are shown in Table 1, Correlation Analysis Class of vehicles' correlation to gas prices.

Table 1: Correlation Analysis

Class of vehicles correlation to gas prices

Body-on-Frame SUV to gas prices

Unibody (crossover) SUV to gas prices

Net Difference in Sales between groups and gas prices

These correlations were completed using the function available in Microsoft Excel 2003.

Key findings from the correlational analysis support the forecasting hypothesis that gas prices are actually increasing the sales of Unibody SUVs (.32 correlation) faster than they are leading to the slower sales of Body-on-Frame SUVs (.18 correlation). The results of this analysis say that Body-on-Frame SUVs are being influenced much more significantly than other factors over and above gas price.

To see if the differences between each category of vehicles in terms of monthly shipments are explained by the price of gasoline month-to-month the differences in shipments between each category was derived and then correlated to monthly gas prices. The weak correlation of -.14 shows that the variation in sales between each class of SUV is not influenced by gas in any statistically significant way; clearly there is more going on between each of these segments than just gas prices. In terms of forecasting sales then, the scope of data needs to be expanded to see what additional time periods can provide in terms of insights into the decline of Body-on-Frame SUVs relative to the growth of Unibody (crossover) SUVs. For this analysis the Phase 2 of the forecasting project has been completed.

Phase 2 Analysis

With gas prices in the short-term shown to be more of a predictor of the growth of unibody (crossover) SUVs than a deflator and negative influence on Body-on-Frame SUVs, the next step in defining a forecast for both classes of vehicles is to look at the interrelationships each of these variables have with each other, keeping gas prices constant. The results for fourteen months of sales analysis yield an interesting series of forecasting curves, with the key message that unibody (crossover) SUVs have well established themselves as the next generation of mainstream SUV class of vehicles. The following chart, Unibody Midsize SUV Sales as a % of total SUV Sales, shows the rapid increase in the sales of these newest class of SUVs as a percentage of all midsize SUV sales. From 46% in June 2005 to 54% in July, 2006, the jump of 8% even with gas prices reaching their highest levels ever in the U.S., signals more than just fuel economics influencing SUV sales; there is a fundamental shift occurring in the market.

When a 3-month moving average applied to the unibody (crossover) SUV sales for the last fourteen months the variations on a per monthly basis are smoothed, yet the trending is still showing this class of vehicle being over 50% of SUV sales. The chart, Unibody midsize SUV Sales a % of total SUV Sales (3-Month Moving Average Included) shows the influence of this trending technique.

The most critical aspect of the Phase 2 hypothesis is that the influence of unibody (crossover) SUVs is much more of a significant influence on Body-on-Frame SUV sales than fuel. The three-month moving average shows preliminary trending up of crossover sales.

Widening the time horizon to include 2001 to 2006 shows the significant rise in unibody SUVs since the beginning of the decade. While this chart does not show is the many product introductions in the unibody (crossover) SUV marketplace, yet the rise in sales of these vehicles is unmistakable in the data shown in the chart, Comparing Body on Frame vs. Cross-over SUV Demand 2001-2006 YTD. The data table for unibody (crossover) SUVs shows the rapid ramp of the import manufacturers including Honda, Hyundai, Nissan, and Toyota. Table 2, Unibody SUVs, based on R.L. Polk registration data are shown below.

Table 2: Unibody SUV Sales 2001-2006 (U.S. registrations)

Unibody SUVs (RL Polk Data)

CHEVROLET - EQUINOX

FORD - FREESTYLE

HONDA - PILOT

HYUNDAI - SANTA FE

NISSAN - MURANO

SUBARU - B9 TRIBECA

TOYOTA - HIGHLANDER

Totals by Year

Going back to our Phase 2 hypothesis of unibody (crossover) SUVs being a more statistically significant predictor of Body-on-Frame SUV demand long-term than gas prices, the chart Comparing Body on Frame vs. Cross-over SUV Demand 2001-2006 YTD graphically shows this, yet to prove the hypothesis statistically and also provide a foundation for forecasting demand, statistical significance needs to be established.

Using SPSS Version 13 to produce Kendall's tau_b and Spearman's rho correlation coefficients between the years of 2001 to 2006 results in the strongest correlations between Body-on-Frame and unibody (crossover) SUVs found in Phase 2 forecast testing. With.573 correlation of Unibody directly influencing Body-on-Frame sales in the years sampled. Table 3 provides the results of the query made in SPSS Version 13.

Table 3: SPSS Correlation Coefficients

Kendall's tau_b

BodyOnFrame

Correlation Coefficient

Sig. (2-tailed)

UnibodyCrossover

Correlation Coefficient

Sig. (2-tailed)

Spearman's rho

BodyOnFrame

Correlation Coefficient

Sig. (2-tailed)

UnibodyCrossover

Correlation Coefficient

Sig. (2-tailed)

With the statistical analysis showing reasonably strong predictability, the next step is to evaluate the specific 14-month time series for greater insights into the variability and predictability of the data. What emerges from completing a Linear Regression along with every exponential smoothing techniques for curve fitting is further evidence of linear (.702 regression) in addition to quadratic (.737 regression) shows that variations in Body-on-Frame demand are explained through these statistical techniques.

Table 4: Model Summary and Parameter Estimates

Dependent Variable: BodyOnFrame

Equation

Model Summary

Square

Linear

Logarithmic

Inverse

Quadratic