1. Problem recognition, definition and evaluation
Steel is primarily used for major material in oil and gas projects. The volatility of steel price has therefore impacted actual projects cost overruns. Figure 1 shows the steel prices has shown significant volatility for the last 5 years benchmarked to Gold Price. Hence, the future cost estimation has to consider this positive trend of steel price and use a new alternative currency reference since US Dollar value has declined significantly in recent years as shown in Figure 2.
Figure 1. World Average Steel and Gold Price Jan 96 – Jul 12
Figure 2. US Dollar Index Chart: Jan 1973- Aug 2012
2. Development of the feasible alternative
Gold is used as reference commodity in estimating steel prices particularly for billet and scrap steel. Billet and scrap steel price was analyzed using gold equivalency in week 8 and week 11. In this chapter, gold price forecast model will be developed based on historical data using regression analysis.
3. Development of the outcome
Gold price is analyzed using regression method based on monthly average gold price from January 2000 to September 2012 obtained from World Gold Council (London PM Fix Price), and then forecasted for next three years up to December 2016. Figure 13 describes gold price has exponentially moved for last ten years. Therefore, the gold price estimation model will observe the data starting from January 2000 up to September 2012 in order to picture the latest trend.
Figure 3. Trend Line of Lending Rate against PP USD over Gold
4. Criteria of Selection
Regression model will select the best fit model based on r-squared value as shown in Figure 4 that results Polynomial 5th order is the model that has the highest r-squared value.
Figure 4: Gold Price Forecast using Regression Models (source: author)
5. Analysis
Although Poly 5th order has the highest r-squared value, the forward curve estimates the gold price will decline until negative value that is impossible to happen. Hence Polynomial 4th order is then selected as the best model since it has r-squared 0.9825 as the second high one among others.
6. Selection of alternative
Poly 4th order model is selected as the best model, and it has represented 98.25% of the observed data. To sum up, Equation 1 is used to forecast average monthly gold price until 2016, in which y is Gold Price and x is monthly time unit 153 + n.
y = -2E-06x4 + 0.0008x3 - 0.0229x2 + 1.7963x + 259.1
Equation 1: Average Monthly Gold Price Forecast 2012-2016 (source: author)
7. Performance monitoring and post-evaluation of results
Before applying the resulted formula into forecast model, it has to be tested analysis to describe the reliability of the predictive model. The testing method and gold price forecast will be discussed in the next chapter.
References:
· Steel on the net. (2012, Sep 24). Historical Steel Price. Retrieved from: http://www.steelonthenet.com/pricing_history.php
· US Federal Reserve. (2012, Sep 24). US Dollar Index. Retrieved from: http://www.federalreserve.gov/releases/h10/summary/indexn_m.htm
· World Gold Council. (2012, Sep 24). Interactive gold price chart and downloads. Retrieved from: http://www.gold.org/download/value/stats/statistics/xls/gold_prices.xls
AWESOME, Trian!!! NOW you are producing R2 values which are RELIABLE, ACCURATE and VALID. http://webs.mn.catholic.edu.au/physics/emery/measurement.htm#Accuracy (Be sure in your paper that you put one paragraph in explaining the differences between these three terms!!)
ReplyDeleteWhat you MIGHT want to consider would be take the 2nd Order as the WORST CASE scenario (inflation goes crazy); the 5th Order as the MOST LIKELY scenario and the Linear as the BEST CASE scenario, then project into the future using these three curves?
Now the only remaining task is to take the three types of steel, convert them into ounces of gold and using the numbers from Figure 4 above, project the price of the three types of steel into the future and BINGO!!! Your paper is done...
BR,
Dr. PDG, Jakarta