While we feel comfortable with these figures, it is also required to look at last year s production when trying to predict the price for an April purchase. If you look at Figure A3.2, you can see that price is most dependent on the previous years production and the current inventory level. The regression model gives us a 0.8 R squared. This is because in April, before the next years harvest, a buyer is still buying from the previous years crop. A look at the year 2000 s production plus 2001 beginning inventory shows that supply increased by 5.68%. Using this information we would have:
Table 2
Supply % Change Demand % Change
5.68% 7%
This gives us a change of 4.28% in price based on 5.68 (7*.2). Using this year s price on January 25 of $67.85 cents/lb and decreasing this by the 4.28%, we come up with a price of $64.95 cents/lb.
Now these two prices bring us to our biggest dilemma, which one to use. As previously mentioned, prices in April seem to be based off of last year s price, but it is still affected by what the predicted harvest of that year will bring. So for simplicity sake, we decided to take the average of the two prices for the fundamental price. The average price is $65.50 cents/lb, so the fundamental price analysis for coffee has predicted a price of $65.50 cents/lb for April 25th.
Technical Analysis
The first step in the technical analysis was to graph the cash price data from the past ten years in order help identify which forecasting model to use by looking at the pattern of the data. The graph can be seen in Figure A4.1. The graph shows no noticeable trend. A TSN analysis was done next to confirm whether or not there was a trend. The results from the TSN analysis can be seen in Table A4.2. Since the results of the TSN analysis show over twenty percent noise, a less complex forecasting model should be used since a large portion of the variation cannot be explained.
However, we were not comfortable with the results of the analysis since a large percentage was a trend. We then performed a TSN analysis for the past three years to see if it would give us different results. However, the analysis still gave us a large percentage of noise. This narrowed down the forecasting method possibilities to a simple moving average, a weighted moving average, or the na ve method.
The results from the averages and na ve method are in Table A4.3. The na ve method is best since it produced the lowest overall error. The two-month weighted moving average error was very close to the na ve error. However, the two month weighted moving average was almost a na ve forecast since the weights of .9 (w1) and .1(w2) produced the best results. To confirm that the na ve model would be best, we also analyzed data from only the past three years using the average based models and the na ve method. The na ve method still proved to be the best model.
We believe that the na ve model was the best method because of the uncertainty of the variation, the nature of the market for the commodity, and the time frame. The variation of the coffee supply and price depend on the weather, disease, natural disasters, and government regulations. Therefore, a trend pattern is not likely. The prices in the cash market for coffee have been relatively stable in recent years due to the consistency of the market conditions. Lastly, since we only are forecasting four months in advance, the price is less likely to dramatically fluctuate month to month, which makes the na ve model best. Using the na ve method, our forecasted price for April 25th is $64.39 cents/lb.
Recommendations
The first concern with our recommendation of when to buy coffee centers on which price to use, the fundamental price or the technical price. Our group decided to average out the fundamental and the technical price. The technical we know will have some error because it is na ve and the April price is the forecasted January price. The fundamental on the other hand has better accuracy according to the trend analysis and R squared value, but with demand numbers only estimated, these numbers can be somewhat risky. The forecasted price that our group came up with is equal to 64.95 cents/lb on April 25th. By using this forecasted price, three different options can be looked at to determine when we want to buy the coffee and if a futures contracts will be necessary. All three options are illustrated on Figure A5, so not all values will be addressed in the report
Option number one dealt with buying in the cash market on January 25th at 67.85 cents/lb. In order to do option one, we needed to convert our figures because our price was in cents/lb and buying options in the cash market are only sold in 60 kilograms bags. The conversion mark for lbs to kilograms is 2.2, so we first divided the $500,000 allotted to buy coffee by the price per pound to get a total number of pounds that will be bought. The second step was to take the total pounds and divide those by the 2.2 to come up with how many kilograms $500,000 can buy. This number was then divided by 60, because this is the amount of kilograms that the bags are sold in and came up with a total of 5,583 60Kg bags that we are going to purchase. With these numbers we were then able to find the total price for option number one by taking the $500,000 spent and multiplying it by the 2% holding cost per month times three months and came up with a total cost of $530,000 dollars to purchase 5,583 bags of coffee.
Option two deals with buying the cash product on April 25th. Since our group has predicted a price drop in the cash market for coffee, this has a lower price than option one. With our forecasted price of .6495 cents/lb, it would cost $478,629.54 to purchase 5,583 bags of coffee using option number two.
The third option is to buy in the cash market on April 25th and to buy a May Futures contract on January 25th and sell it back on April 25th. This is referred to as long hedging. By having the cash price drop 2.90 cents/lb using the Jan. 25th price minus our forecasted April 25th price, we would save $21,370.68. If we bought a futures contract on Jan. 25th for 70.55 cents/lb and assumed that it would be worth our forecasted cash price of 64.95 cents/lb on April 25th, then you would lose $41,267.52 on the futures contract. The total loss through hedging would be $19,896.84 plus a $2,500 commission for the purchase and selling of the futures contract and you have a total cost of $522,396.84 to purchase the 5,583 60Kg bags of coffee.
As you can see, option number two would be the recommended option, as it is nearly $44,000 cheaper than the next best option, option number three. In order for option number three to be better, coffee would have to rise to $.702 cents/lb, which was found by using breakeven analysis. Based on the standard deviation of .29 cents/lb from the past three years of cash price data, the probability of coffee rising to $.702 appears to be great in April. However, we are very confident with our forecast and are going to stay with option number two. We strongly believe that the price of coffee will fall, not rise.