Ted Ralley is working on conducting a forecast for the upcoming year for an automobile part company. The data that will be used for this research has been collected from the quarterly sales from the previous four years. Ted wants to determine what is most accurate way to determine the forecast for 2008. The model should also help determined if the economic situation and oil prices are affecting significantly the sales of the company. The two models that were provided were thoroughly analyzed to determine which model was the most appropriate to utilize. These models were a regression model with factors, seasons and an additive Holt-Winters model. The forecasts also show that there is a significant change in the sales with the economic hardship and oil prices. It was concluded that the Regression with Econometric Variables would be the best method to use to forecast the sales for 2008, estimating a 255,927,955 for that year.
With the economy continuously deteriorating everyone seems to be getting hurt financially, even the automotive industry, which has deepening the economic recession. Automotive part suppliers continued to experience heavy debt and overcapacity caused by production cuts by automakers, specifically including the big 3 (Ford Motor Company, General Motors and Chrysler). The suppliers are also being pressed by higher energy and input materials’ costs. It has been determined by Industry analyst that automotive companies that accounted for more than $72 billion in sales have filed for chapter 11 protections in 2008. The number of Bankruptcies will continue to rise as the years go by. Domestically, Losing the big 3 to U.S affiliates of foreign- based manufacturers and imports in 2008 have caused a dramatic 50% drop in the market share.
Most US suppliers are dependent on these three companies aforementioned. U.S suppliers are currently facing the challenge of penetrating automakers’ supply chains, mostly because these relationships have been long-established with home-market supplies. Ted Ralley is the director of a marketing research for a manufacturer of spare automobiles parts and it’s working on conducting a forecast for the upcoming year. Ted is aware of the forecasting errors and how costly they can be which is why these numbers must be as accurate as possible. In order to perform this forecast, Ted has collected the data on quarterly sales for the previous four years and ran several forecasts using time series forecasting methods. Ted noticed that economic activity and oil prices have impacted significantly the auto part sales and decided that the forecast will be more accurate using econometric variables. Problem
Will the econometric variables be a better predictor of sales for the coming year, given the current economic activity and oil prices? Analysis
This analysis consisted of the evaluation of the regression model with factors, seasons and the additive Holt-Winters method to generate an accurate forecast of how econometric variables have affected the Auto Parts industry. The analysis involved calculating the errors metrics for the three models (mean absolute percentage error (MAPE), root mean square error (RMSE), MAPE and Theils’ U-statistics (U)) and comparing them against each other. The error metrics were calculated by using the formulas shown below: Table 1.1 Error Metrics Formulas:
After studying the data provided it could be determined that there is an upward trend with obvious seasonality. Another factor that played a role in these regressions was the removal of the first two years in order to meet Holt-Winters method guidelines. The first regression was conducted using Factors was generated by utilizing the data that provided by Ted Ralley from a large manufacturer of spare auto parts for automobiles. The data consisting of the quarterly sales for the previous four years was the dependent variables and independent variables consisted of Time, quarter 2, quarter 3, quarter 4. In this regression quarter 1 was removed in order to avoid over forecasting and binary coding was used to generate dummy factors. After the regression was completed, the independent variables were tested to determine their significance, which was done by performing a regression on the data through Microsoft Excel. Quarter 4 was removed from the model due to the fact that it was statistically insignificant. This was determined by using backward elimination, which means, a variable that has a P-Value that is greater than .05, is considered insignificant and should be removed from the data and a new regression should be completed.
The results from the new regression, shown below, have a P-Value less than .05 being sufficient to reject the null hypothesis (Ha). A very strong positive linear correlation between sales and all the independent variables combined with a 95.47%, leaving an unexplained variance of 4.53 is also demonstrated. According to the textbook “the most common measure of overall fit is the coefficient of determination (R2)”. Another important measure is the “standard error (Se), which is derived from the sum of squared residuals for n observations and k predictors” (Poane, Seward, 2013). A smaller Se Indicates a better fit, in this case the Se will be off by around 3.9 million. The coefficients used to run the forecast for 2008 are the following: intercept coefficient + coefficient time x time 1 plus coefficient q2* code for Q2 dummy variable for q2 + plus coefficient q3. Square error was used to find the magnitude of the error; the absolute value of the error to the sales was found and then preceded to calculate to numerator. Numerator and denominator will be calculated in other to use Thiels’ U. Numerator was calculated as follow: difference between sales minus the sale of initial sale (difference q1-2 sales) /divided by q1 and squared.
Poane, D., & Seward, L. E. (2013). Business Modeling Customized Readings for QNT5040. : Mc Graw Hill Education.
Microsoft Office Excel. (2007). Redmond, WA: Microsoft Corporation.
Albright, Winston & Zappe (2010). Business Modeling, Selections from 4e – QNT 5040 (4th ed.). Mason: Cengage Learning. Aczel,A & Sounderpandian,J (2009). Complete Business Statistics 7th edition (592). : Mc Graw Hill Education.
U.S. Automotive Parts Industry Annual Assessment. (2009, April 1). . Retrieved June 6, 2014, from http:[email protected]_oaai/documents/webcontent/tg_oaai_003759.pdf