ACE 427 Spring 2013 Lecture 5 Forecasting Crop Prices Using Fundamental Analysis: Ending Stock Models by Professor Scott H. Irwin Required Reading: Westcott, P.C. and L.A. Hoffman. Price Determination for Corn and Wheat: The Role of Market Factors and Government Programs. US Department of Agriculture, Economic Research Service, 1999. (427 class website) Abbott, P. C., C. Hurt, and W.E. Tyner. What s Driving Food Prices in 2011? Issues Report, The Farm Foundation, July 2011. (427 class website) ACE 427, University of Illinois 5-1
Fundamental Analysis Definition: An assessment of based on the underlying and factors and the changes in those relationships Motivated by economic of supply and demand The task of the market is to establish a price that will Fundamental analysis can be thought of as the process of anticipating the market clearing price Techniques: Subjective judgment to sophisticated statistical models Goal: Estimate and compare to Bullish: Value > Price Bearish: Value < Price ACE 427, University of Illinois 5-2
Ending Stocks and Price Ending stocks indicate the between supply and demand Ending stocks, price Ending stocks, price Relationship between ending stocks and price is often used to forecast prices 6,000 5,000 US Corn, Farm Price and Ending Stocks, 1975/76-2011/12 7.00 6.00 Ending Stocks (mil. bu.) 4,000 Price 5.00 3,000 4.00 2,000 3.00 1,000 2.00 Stocks 0 1.00 1975/76 1981/82 1987/88 1993/94 1999/00 2005/06 2011/12 Price ($/bu.) Source: USDA Marketing Year ACE 427, University of Illinois 5-3
US Corn, Ending Stocks, 1975/76-2011/12 6,000 5,000 Million Bushels 4,000 3,000 2,000 1,000 0 1975/76 1982/83 1989/90 1996/97 2003/04 2010/11 Source: USDA Marketing Year US Corn, Total Use, 1975/76-2011/12 14,000 Total Use (million bushels) 13,000 12,000 11,000 10,000 9,000 8,000 7,000 6,000 5,000 1975/76 1982/83 1989/90 1996/97 2003/04 2010/11 Source: USDA Marketing Year ACE 427, University of Illinois 5-4
70 US Corn, Ending Stocks/Total Use, 1975/76-2011/12 60 Ending Stocks/Use (%) 50 40 30 20 10 0 1975/76 1982/83 1989/90 1996/97 2003/04 2010/11 Source: USDA Marketing Year ACE 427, University of Illinois 5-5
Building an Economic Model A model is an from the real world Must be yet capture economic relationships A model of a market can be thought of as one or more that describe the important among the variables in the market The Simplest Market Model QD t = a b P t QS t = d + e P t QD t = QS t (Demand) (Supply) (Equilibrium) ACE 427, University of Illinois 5-6
Economic Model Underlying Balance Sheets before Planting Price Supply P 0 Demand Q 0 Quantity ACE 427, University of Illinois 5-7
Adding Shifter Variables In the simple model, there is only one because nothing ever! In reality, we know that: Demand curves shift due to changes in the, and other variables Supply curves shift due to changes in the, and other variables Key point: Changing equilibrium prices and quantities are driven by changes in the level of ACE 427, University of Illinois 5-8
A More Realistic Market Model QD t = a b P t + c I t (Demand) QS t = d + e P t QD t = QS t (Supply) (Equilibrium) Price P2 t QS t P1 t QD t (I t level 2) QD t (I t level 1) Q1 t Q2 t Quantity ACE 427, University of Illinois 5-9
An Even More Realistic Market Model QD t = a b P t + c I t (Demand) QS t = d + e P t - f F t (Supply) QD t = QS t (Equilibrium) Price P2 t QS t (F t level 2) QS t (F t level 1) P1 t QD t (I t level 2) QD t (I t level 1) Q1 t Q2 t Quantity ACE 427, University of Illinois 5-10
A Complete Pricing Model in Linear Form (1) S t = a 1 + a 2 p t-1 a 3 F t Supply (2) D t = b 1 b 2 p t-1 + b 3 I t Domestic/Export Demand (3) K t = c 1 c 2 p t c 3 i t Stock Demand (4) S t = D t + K t Equilibrium Condition Forecasting with Pricing Model In equilibrium, the relationship between prices and ending stocks can be found by the stock demand function and fixing the value for the shifter variable in this equation (3): p t = [(c 1 / c 2 ) + (c 3 / c 2 ) i t ] (1/c 2 ) K t Basic pricing function used by many market analysts Essential to note that this assumes shifter variables are held! ACE 427, University of Illinois 5-11
Price ($/bu.) 6.50 6.00 5.50 5.00 4.50 4.00 3.50 3.00 2.50 2.00 1.50 US Corn, Farm Price vs. Ending Stocks/Total Use, 1975/76-2011/12 0 10 20 30 40 50 60 70 Source: USDA Stocks/ Use (%) US Corn, Farm Price vs. Ending Stocks/Total Use, 1975/76-2011/12 Price ($/bu.) 6.50 6.00 5.50 5.00 4.50 4.00 3.50 3.00 2.50 2.00 1.50 y = -0.0269x + 3.236 R² = 0.1702 0 10 20 30 40 50 60 70 Source: USDA Stocks/Use (%) ACE 427, University of Illinois 5-12
Logical Characteristics of Relationship between Price and Stocks As ending stocks approach, theoretically, there is no for price As ending stocks get very large, price is unlikely to go below some minimum level Theoretical Functional Form between Price and Ending Stocks Price Ending Stocks/Use ACE 427, University of Illinois 5-13
Transformations and Least Squares Regression All is not lost when the relationship between x and y is LS regression works for any non-linear transformation of the We can take logs, divide or multiply variables Valid as long as we do not transform (e.g. square, cube, etc.) Regression Functional Forms between Price and Ending Stocks That Account for Non-Linearity Double-log functional form: ln(y) = b 1 + b 2 ln(x) Reciprocal functional form: y = b 1 + b 2 (1/x) ACE 427, University of Illinois 5-14
Properties of the Reciprocal Functional Form y = b 1 + b 2 (1/x) b 1 measures the (or maximum) level of y b 2 does not measure change in y for a one-unit change in x, but instead change for a one unit change in 1/x Hence, the slope (change in y for a one unit change in x) for different of x y/ x = dy/dx = -b 2 (1/x 2 ) ACE 427, University of Illinois 5-15
Corn: Setting Up the Data for a Reciprocal Model in Excel Total Ending x=stocks/ y=corn Year Use Stocks Use (%) 1/x Price 1 1975/76 5767 633 11.0 0.09 2.54 2 1976/77 5789 1136 19.6 0.05 2.15 3 1977/78 6207 1436 23.1 0.04 2.02 4 1978/79 6995 1710 24.4 0.04 2.25 5 1979/80 7604 2034 26.7 0.04 2.48 6 1980/81 7282 1392 19.1 0.05 3.12 7 1981/82 6975 2537 36.4 0.03 2.47 8 1982/83 7249 3523 48.6 0.02 2.55 9 1983/84 6693 1006 15.0 0.07 3.21 10 1984/85 7032 1648 23.4 0.04 2.63 Corn Price ($/bu.) US Corn, Farm Price vs. Ending Stocks/Total Use, 1975/76-2011/12 6.50 6.00 5.50 5.00 4.50 4.00 3.50 3.00 2.50 2.00 y=14.78(1/x) + 1.71 R2=0.31 1.50 0.0 10.0 20.0 30.0 40.0 50.0 60.0 70.0 Source: USDA Ending Stocks/Use (%) ACE 427, University of Illinois 5-16
Different Approaches to Account for Shifts in Relationship Include variables directly in the pricing model and estimate model for the entire sample period Approach taken by Westcott and Hoffman 3 additional independent variables for corn 5 additional independent variables for wheat Estimate pricing models for Shifter variables have different levels across the sub-periods The level of shifter variables is assumed to be relatively within a subperiod ACE 427, University of Illinois 5-17
US Corn, Farm Price vs. Ending Stocks/Total Use, 1990/91-2005/06 3.50 Corn Price ($/bu.) 3.00 2.50 2.00 y=8.34(1/x) + 1.64 R2=0.78 1.50 0.0 5.0 10.0 15.0 20.0 25.0 30.0 Source: USDA Ending Stocks/Use (%) Slope changes with x Ending stocks/use = 5 y/ x = -8.34 (1/5 2 )= -$0.33/bu. Ending stocks/use = 25 y/ x = -8.34 (1/25 2 )= -$0.01/bu. ACE 427, University of Illinois 5-18
Corn Price ($/bu.) US Corn, Farm Price vs. Ending Stocks/Total Use, 1990/91-2011/12 6.50 6.00 2011/12 5.50 5.00 2010/11 4.50 2007/08 4.00 2008/09 3.50 2009/10 y=8.34(1/x) + 1.64 3.00 2006/07 R2=0.78 2.50 2.00 1.50 0.0 5.0 10.0 15.0 20.0 25.0 30.0 Source: USDA Ending Stocks/Use (%) ACE 427, University of Illinois 5-19
What Changed During the 2006/07-2011/12 Marketing Years? All else equal, supply shifted to the High input prices due to rising energy prices? Weather? Or, demand shifted to the Ethanol demand? Export demand? Feed demand? Stock demand? Some of supply and demand shifts ACE 427, University of Illinois 5-20
Building a New Model Assume that 2006/07 is a transition year and remove it from the sample Add a dummy variable to the regression to allow the intercept for the model to increase during 2007/08-2011/12 (D is zero for 1990/01 to 2005/06 and one otherwise) Corn Price ($/bu.) 6.50 6.00 5.50 5.00 4.50 4.00 3.50 US Corn, Farm Price vs. Ending Stocks/Total Use, 1990/91-2011/12 Estimated model: y=11.57(1/x) + 1.40 + 2.16D R2=0.89 2007/08-2010/11: y=11.57(1/x) + 3.56 3.00 1990/91-2005/06: 2.50 y=11.57(1/x) + 1.40 2.00 1.50 0.0 5.0 10.0 15.0 20.0 25.0 30.0 Source: USDA Ending Stocks/Use (%) ACE 427, University of Illinois 5-21
First ACE 427 Corn Balance Sheet Estimates for 2013/14 USDA/WASDE USDA/WASDE ACE 427 Item 2011/12 2012/13 2013/14 Supply: Planted Acreage (million acres) 91.9 97.2 98 Harvested Acreage (million acres) 84.0 87.4 91 Yield (Bushels / Acre) 147.2 123.4 160 Beginning Stocks (million bushels) 1,128 989 632 Total Production (million bushels) 12,360 10,780 14,528 Imports (million bushels) 28 200 5 Total Supply (million bushels) 13,516 11,969 15,165 Consumption: Feed and Residual (million bushels) 4,547 4,350 4,900 Food, Seed, and Industrial (million bushels) 6,437 5,887 6,150 Ethanol (million bushels) 5,021 5,000 4,700 Exports (million bushels) 1,543 900 1,450 Total Consumption (million bushels) 12,527 11,237 12,750 Ending Stocks (million bushels) 989 632 2,415 Ending Stocks/Total Consumption (%) 7.9% 5.6% 18.9% Average Price ($/bu.) 6.22 7.20??? Note: USDA WASDE estimates were released in February 2013 ACE 427, University of Illinois 5-22
Ending Stocks/Use Forecast: (2,415/12,750)*100 = 18.9% Forecast price: y = 11.57 (1/18.9) + 3.56 y = 11.57 (0.053) + 3.56 y = $4.17/bu. Corn Price ($/bu.) 6.50 6.00 5.50 5.00 4.50 4.00 3.50 US Corn, Farm Price vs. Ending Stocks/Total Use, 1990/91-2011/12 $4.17 Estimated model: y=11.57(1/x) + 1.40 + 2.16D R2=0.89 2007/08-2010/11: y=11.57(1/x) + 3.56 3.00 1990/91-2005/06: 2.50 y=11.57(1/x) + 1.40 2.00 1.50 18.9% 0.0 5.0 10.0 15.0 20.0 25.0 30.0 Source: USDA Ending Stocks/Use (%) ACE 427, University of Illinois 5-23
Final Thoughts We must always keep in mind that price and ending stocks are determined The true relationship is: Supply and Demand Shifter Variables Price and Ending Stocks In other words: It is not logical to specify the level of without knowing It is not logical to specify without knowing the level of Therefore, price and ending stocks models are only a or to forecasting crop prices Knowledge of underlying relationships is required to make more accurate price forecasts ACE 427, University of Illinois 5-24
In addition, forecasts from price and ending stocks models for or stock levels may be quite sensitive to the assumed form Reciproal Double-log Log-linear Bottom line: price and ending stock models may be a good starting point, but they should be used with a great deal of caution For a more in-depth treatment of these issues, I highly recommend: William G. Tomek and Kenneth L. Robinson. Agricultural Product Prices, Fourth Edition. Cornell University Press: Ithaca, NY, 2003. ACE 427, University of Illinois 5-25