Using data gathered from the USDA databases, this research will estimate the domestic US demand for corn. All price data are reported in 1983 dollars using the consumer price index published by the Bureau of Labor Statistics. An econometric model will be estimated with two-stage least squares (TSLS).
The TSLS estimation method was chosen due to the fact that a simple multiple regression would fail to capture variations in the regressor that are correlated with the error term. In the supply and demand framework, time series equilibrium points represent movements in both the demand and the supply curves. To isolate the movements in the endogenous variables that are uncorrelated with the error term, a set of instrument or exogenous variables must be used to overcome the identification problem. In essence, we are extrapolating variation of corn demand that, in using ordinary least squares, would otherwise be lost within the error term, out into the open via the use of instrument variables. The use of TSLS in this manner is convenient because it allows for the use of a single estimation procedure, while utilizing information from both the demand and supply curves. By using this method, a clearer picture of corn demand can be had by isolating and using information from both the demand and supply of corn. Below is a brief exposition on the TSLS estimation method following Stock and Watson (2006).
Consider the bivariate linear equation,
where Yi is the dependent variable, X i is the independent variable, and µi is the error term representing omitted factors that determine Yi. If correlation exists between Xi and the error term, ordinary least squares would result in inconsistent estimates of the parameters a1 and ß1. The instrumental variable method — in this case, TSLS— uses exogenous variable Zi to isolate that portion of Xi that is uncorrelated with µi . The first “stage” decomposes the independent variable Xi into two components, one that is correlated with the regression error µi, and another that is uncorrelated with the error term. This first stage begins with a population regression linking Xi and Zi:
where p0 is the intercept, p1 is the slope, and vi is the error term. This gives the part of Xi, p0 + p1Zi, which can be predicted by Zi. This portion of Xi is uncorrelated with the error term µi in equation 3.1 because Zi is exogenous. The other component of Xi is vi, which is the portion of Xi that is correlated with the original error µi. The first stage of TSLS uses ordinary least squares (OLS) to estimate the parameters p0 and pi. The second stage then estimates the dependent variable with OLS,
using the estimated Xi, denoted Xi, and disregarding the error term vi.
For the purposes of the corn demand model in this research, the instrumental method is used to identify the equilibrium points by using exogenous information to separate the movements in both supply and demand. In the estimation, the log of all variables was used in accordance with general econometric and statistical practices. The following sections discuss the determinants of supply and demand for corn.
3.1 Demand SideThe demand equation to be estimated is given in the following equation, and summarized in Table 3.1.1:
Table 3.1.1 Corn Demand
|
Quantity corn produced (millions of bushels) |
The demand for corn is assumed to have three distinct origins: that from livestock producers, ethanol producers and human consumers. According to the ERS Feed Grains Database, of the 10.5 billion bushels of corn utilized domestically in the US during the 2007 market year, more than 55% (5.95 billion bushels) was used as feed for livestock. Cattle feed is comprised of a mix of 11% crude protein mixed with feed grains, typically corn (“Agricultural Alternatives”). The link between livestock markets and the corn market will be represented in the model by the prices of livestock production inputs. Of the corn used to feed livestock in Iowa in 2005/2006, approximately 53% went to hogs, 34% to beef cattle, and 12% to poultry (“Corn Use”). These percentages are used to weight a summation of the prices to create an index (PLVI) that captures the relative effects of each industry’s price on the quantity demanded of corn. This price index is hypothesized to show a positive relationship with the price of corn.
According to the ERS Feed Grains Database, in 2007, 3.2 billion bushels, or more than 30% of the corn produced domestically, was used to produce fuel ethanol. This percentage has grown rapidly over the past decade, and was forecasted by the ¬¬USDA to comprise 20% of corn consumption in the US in the 2006/2007 season, which in fact fell short of the true value (Hoffman et al 2007). Thus it would seem reasonable to assume that if the price of ethanol increases due to increased demand, the price of corn would then increase as the demand for corn increases.
Ferris and Joshi (2004), the production of ethanol should be expected to increase due to a combination of the reduction of MTBE as a blend, federal renewable fuel standards, and tighter restrictions on air quality. And as seen in Tokgoz and Elobeid (2006), ethanol has a complementary relationship with gasoline in the US due to its predominant use as a blend rather than a stand alone fuel. These observations about ethanol and its relationship with corn support the inclusion of the price of ethanol (PEN) in the demand for corn; as the production of gasoline and refining of oil continues to grow, so will the production of ethanol. The price of ethanol is not determined entirely by the free market, however, because of government policies.
The price of ethanol is in fact subsidized in several ways. The first comes from direct government intervention in domestic ethanol markets. Such intervention occurs in the form of mandated production levels. The Clean Air Act of 1990, for example, mandates a certain level of oxygenated fuel in areas with air quality issues. Ethanol production is also subsidized through tax credits via the Energy Tax Act of 1978, which introduced a $.40 per gallon motor fuel excise tax exemption to ethanol blends of at least 10 percent by volume. Currently, due to several tax laws that have since been adopted, the tax credit per gallon stands at $.51 through 2010 (Elobeid and Tokgoz 2006). The third way in which ethanol prices are subsidized is more indirect. US trade policy on ethanol includes an ad valorem tariff of 2.5 percent in addition to an import duty of $.51 per gallon (Elobeid and Tokgoz 2006). In the following sections, it is important to note that the combination of mandated production, tax credits, and protection from international prices all influence the equilibrium price of ethanol.
The last portion of corn demand in the model is human consumption. This portion includes corn used in the production of high fructose corn syrup, corn starch, corn sweeteners, cereal or other food products, and beverage alcohol. The amount of corn used in human consumption in 2007 amounted to slightly more than 12% of total corn production (ERS, Feed Grains Database 2008). Sugarcane is a viable substitute for corn as a sweetener in human consumption. For this reason, the price of sugarcane (PSC) is included in the model of corn demand to capture the substitution effect over the years between corn syrup and sugarcane. In fact, according to data from the ERS, the US per-capita use of high fructose corn syrup has been steadily increasing over the years, indicating that it is a competitor of sugar. The preceding observations concerning sugar would suggest that as the price of sugarcane decreases, sugar would be substituted as a sweetener for high fructose corn syrup, decreasing the demanded for corn.
As with ethanol, the price of sugar is hardly a product of free market forces alone. Sugar is one of the most heavily subsidized commodities in the agricultural industry. Sugar markets have been affected by statute since 1789 when the first Congress of the US imposed a tariff on foreign sugar. Acts such as the Sugar Act, also known as the Jones-Costigan Act, solidified government intervention in sugar markets with a series of quotas dictating production (Alvarez and Polopolus 2008). Federal sugar programs in the early 1970s did away with many subsidies to the sugar industry while sugar prices reached record highs. During the mid-1970s, prices once again fell and production costs soared, prompting federal programs to once again introduce price supports and loan programs (Alvarez and Polopolus 2008). Since then, price supports and loan programs have been a consistent part of the sugar industry.
3.2 Supply SideThe supply equation that will support the estimation of corn demand is given in the following equation, and summarized in Table 3.2.1:
In this model, the supply of corn is assumed to be a function of the price of corn, input prices, and the level of government payments. Farm production is inherently dynamic with time lags and expectations dictating current-year production. The harvest in the current year can only be as much as the plantings in the previous year, and thus, many of the production decisions of the farmer take place in the year prior to harvest. Land conversion from one crop to another, or simply expanding a current crop’s plantings, is costly. It is for this reason that the quantity harvested in the previous year (QHK), which implies an area planted, is included in the model of corn supply. The quantity harvested in the previous year captures the switching costs of land in changing crops year over year. Also, farmers can store excess production if prices received are unsatisfactory. Waiting to sell may in turn lead to higher prices received and thus quantity supplied of corn is also a function of year end stocks of corn (QSE). Quantity supplied of corn is assumed to be a function of the cost of inputs to production. Costs associated with fertilizer and seed total almost 60% of operating costs for farmers in 2006 (ERS, Commodity Costs and Returns 2008). For this reason, and in light of the above stated dynamics, the lagged prices of both of these inputs (PFTK, PSDK respectively) were included in the model of supply.
Government assistance programs have been a part of grain markets since the early 1900s. These programs have changed shapes many times and will change again under the legislation currently in Congress. Today, the structure of agricultural subsidies consists of several different entities: direct payments, marketing loans, countercyclical payments, conservation subsidies, insurance, disaster aid, export subsidies, and agricultural research and statistics (Edwards 2008). There exists no number, variable, or data that fully encompasses all of the changing aspects of each of these payments. As a proxy, to capture the effects of government subsidies on the supply of corn, the reported monetary contribution (PCCC) of the Commodity Credit Corporation (CCC) to feed grain producers and the total quantity of CCC corn stocks (QCCC) are included in the model.
The CCC is a branch under the USDA Farm Service Agency that is charged with the responsibility of dispersing government funds to farmers. The reported monetary contribution, or PCCC, is an entry within the “Net Budgetary Expenditures” reported by the CCC that presents the “Total support & related.” This total includes the value of all deficiency payments, production flexibility contracts, loan deficiency payments, marketing loss payments, diversion payments, disaster payments, and storage payments, minus the value of loan repayments, sales proceeds, and other receipts (United States Department of Agriculture 2001). The resulting value, PCCC, is a proxy for the total net governmental value added to the feed grains industry, which includes corn, in the form of subsidies. Quantity of CCC stocks is simply the reported amount of government-owned corn stocks per year. Changes in the CCC’s balance sheet gives the amount of government funds being pushed into the corn industry each year and to the amount of corn held off the market by the CCC, which captures the variations in quantity supplied due to changing government holdings.
© 2007 UNIVERSITY OF CENTRAL FLORIDA UNDERGRADUATE RESEARCH JOURNAL -COMPUTER CENTER II ROOM 209- ORLANDO, FL 32816-0086 - (407) 823-3125
Office of Undergraduate Research
Contact the webmaster with website questions.
View PDF