Following the passage of the 1985 farm bill, the model was used extensively to address farm program implementation issues such as conservation compliance, flex, and marketing loans. It was during this time that AFPC started developing representative crop farms outside of Texas. Presently, AFPC has more than 110 representative farms in 28 states using the panel farm process (Figure 2). FLIPSIM is presently capable of simulating crop farms that also have dairy and livestock (cattle, hogs, poultry, sheep, hair goat, and meat goat ) enterprises. The model has been used to simulate farms in other countries, such as: Canada-dairy, Mexico-dairy and hogs, Kenya-dairy, Philippines-rice, South Africa-grain farms, Brazil and Argentina-cattle, Gamay-dairy, France-crops, and Mali-crops.

Figure 3 illustrates how FLIPSIM simulates the annual activities of a representative farm or ranch using price projections from sector models and assumptions regarding policy options. Information to describe a representative farm is obtained from a panel of producers; the panel is used to validate the model's ability to simulate their representative farm. The model incorporates actual price and production risk faced by the farm by using historical yields and livestock production information. Price projections from sector level models are made stochastic by incorporating historical price risk. Thus the model is capable of simulating representative farms under risky conditions faced by actual farms and ranches.

This document is organized as follows. The first section describes the process that AFPC uses to develop representative farm data used for simulation with FLIPSIM. The second section describes the major activities in a semi-technical manner. FLIPSIM output and key output variables are described in the third section.

• Size of the operation (acres, head, etc.),
• Tenure (acres owned and leased) and asset values,
• Enterprises (crops, livestock, dairy, etc.),
• Costs of production for each enterprise,
• Fixed costs for the overall operation,
• Yields and a history of yields and farm program participation, and
• Machinery complement and replacement strategy.
• Policy History (base acres and payment yields)

The producers who make up the panel are identified by local facilitators (usually county agricultural extension agents) and are representative of a commercial operation in the area. Normally, two farms are developed in each region using separate panels of producers; one is representative of moderate scale full-time farm operations while the second panel represents farms that are two to three times larger.

Each farm panel is interviewed using a consensus building process. Producers are asked to develop a typical farm drawing on their personal operations and experience. During the interview process, the farmer's information is entered into a FLIPSIM input file. Once the interview is complete, the model is run to show the producers the projected income statement, cash flow and balance sheet (pro forma financials) for their representative farm. The producers are then asked to adjust their input values for costs, yields, etc., until the results for the first year match up with recent experiences. Adjustments to debt levels and current market prices are generally made by the panel until they are satisfied that the model can simulate their farm data. This interactive validation process has proven to be helpful to the panel farm process because it gives the producer panel an immediate feedback that their efforts were worthwhile. The representative farms are updated every three years, or as often as the panels want to update their farm data. Most all panel farms have been updated five times and the oldest panel farm has been maintained since 1980.

The model is capable of simulating a farm with 1-20 crop enterprises over a 10 year planning horizon for 100 or 500 iterations. The livestock enterprises that are included in the model are: dairy, cow/calf, cow/calf/retained ownership, stocker steers, beef cattle feedlot, sheep, mohair goats, and meat goats. The model is sufficiently general that it can simulate any or all of these enterprises on any size farm in any state in the United States. All information for a farm is read as input data, as there are no econometric equations used in the model. All equations are either identities or probability distributions.

Annual projections for mean prices, farm program variables, interest rates, rates of inflation, and tax and depreciation provisions are exogenous inputs for FLIPSIM and are represented in the left-hand column of Figure 4. Projections of prices, farm program, and macroeconomic variables come from the FAPRI Baseline, which includes projected values for macroeconomic variables developed by WEFA/DRI. Information to describe a farm for FLIPSIM is represented as the variables in the second column from the left in Figure 4. The multivariate empirical probability distribution for simulating stochastic yields, livestock production variables, and prices is represented at the top of Figure 4 and is developed for each farm using 10 years of actual yields for a panel member's farm and 10 years of annual prices for the region¹.

When FLIPSIM is simulated in the deterministic mode, average yields, livestock production values, and prices are used and the model is run for one iteration. For a stochastic analysis, yields, livestock production variables, and prices are selected at random from the multivariate distribution so the historical correlation between yields and prices is maintained, the historical autocorrelation of prices is maintained, and the relative variability for these variables is constant (or controlled by the user) over time. Output from a stochastic simulation consist of 100 (or 500) values for each of the output variables, i.e., the information to estimate the parameters for probability distributions of key output variables in the model.

Production for each crop i in year t is calculated as the product of the stochastic yield for crop i and harvested acres for crop i (Figure 4). Harvested acres are determined either by the analyst or the model, based on the crop's expected net returns. The model either uses the analysts preset crop mix or solves for the profit maximizing crop mix each year, given agronomic constraints. The model can adjust acres planted from year-to-year based on several formulas for calculating expected net returns and a "reservation change" in net returns per acre required to make a change.

Crop receipts from the market place are the greater of the stochastic price for crop i or its loan rate (if available for the policy being simulated) times the owner/operator's share of production. Government payments (counter cyclical, direct, and loan deficiency payment) are calculated for each crop based on the relevant formula and the policy being simulated. Government payments are appropriately reduced by the landlord's share of the crop on share-rented crop land and the payment limitation specified by the analyst.

FLIPSIM has been expanded to simulate alternative marketing strategies. The marketing section of the model generates stochastic monthly cash and futures prices for each crop. Marketing strategies, such as, sell at harvest, sell different portions of the crop in multiple months after harvest, forward contracting and hedging can be analyzed.

Annual variable production costs are calculated for each crop as the product of harvested acres and the inflation-updated per acre costs for seed, fertilizer, herbicides, insecticides, fuel, irrigation, and the other production costs (Figure 4). Harvest costs are computed as the product of harvested acres, stochastic yield and the inflation-updated harvesting cost per yield unit (e.g., $/bu). Fixed costs for year t-1 are updated using the inflation rate for prices paid for production items to obtain total fixed costs in year t². Operating interest costs are calculated as the product of the operating interest rate, costs of production, and fraction of the year operating debt is used, adjusting for the use of cash reserves to self-financing operating expenses. Interest costs for land and machinery loans are calculated as the product of their respective loan lives, interest rates and outstanding balances. Total cash expense is the sum of variable costs for all crops and livestock enterprises, total fixed costs, and interest costs (Figure 4).

Net cash farm income equals total receipts less total cash expenses (Figure 4). Net cash farm income is used to calculate cash inflows and income taxes. Other components to cash inflows are interest earnings on ending cash reserves (not used for operating purposes) and off-farm income. Annual cash outflows are calculated as the sum of family living expenses, principle payments, federal, state, self-employment, and Medicare taxes, and down payments for machinery replacements (Figure 4).

Annual income taxes (federal and state), self employment taxes, and depreciation are calculated using the 2003 federal income tax provisions, and the individual state income tax provisions for 2003. Annual taxable income is calculated using net cash farm income, depreciation, personal deductions and exemptions, and the current federal and state income tax provisions. Depreciation is calculated for each machinery item (r) in the complement based on the depreciation provisions when the machine was placed into service. At the end of a machine's economic life (year n), it is traded for a replacement. The replacement cost in year n is the machine's inflation adjusted replacement cost, less its market value in year n. The trade in value for a machine in year n is its market value reduced (deflated) for its annual loss in market value.

Annual ending year cash reserves equal total cash inflows less total cash outflows (Figure 4). If cash reserves are negative, the deficit is financed as an operating loan carryover for the next year, thus increasing interest costs in the next year. Ending cash reserves are added to the updated value of land and machinery to calculate total assets at year end. Total liabilities is the sum of real estate, machinery, and livestock debt. Ending net worth is calculated as total assets less total liabilities (Figure 4). A test is made at year end to determine whether the farm is solvent, i.e., if the equity to asset ratio exceeds the analysts minimum of, say, 15 percent. If the farm is solvent, FLIPSIM proceeds to the next year; if not, the model declares the farm insolvent, records the values, and proceeds.

FLIPSIM repeats these annual calculations for each year of the planning horizon. At the end of the planning horizon, the model calculates net present value, present value of ending net worth, and more than 4,000 other key output variables that summarize the iteration. One or five hundred iterations of the planning horizon are run when the farm is simulated in the stochastic mode. After completing the last iteration, summary statistics for the more than 4,000 empirical probability distributions simulated for the farm are computed. The simulated cumulative distribution of net present value for different scenarios can be compared using stochastic dominance or by simply comparing the means, minimums, maximums and standard deviations.

The FLIPSIM schematic in Figure 4 depicts the activities of the model for a crop farm. To simulate a dairy, hog, beef cattle ranch, sheep or goat ranch, the model simply calls into the execution stream additional calculations performed by FLIPSIM for a dairy or livestock enterprise. Livestock consume feed raised on the farm and thus reduce cash receipts from crop sales. Deficit feestocks are purchased based on stochastic prices. The livestock enterprises also produce products that are sold using stochastic market prices (e.g., milk, calves, culled cows, culled bulls, etc.). Cattle and hog purchases required to maintain the cow and sow herd are cash outlays that also offset the balance sheet through the total value of the herd. Fixed values, based on herd replacement information provided by the producers, determine the herd dynamics (i.e., birth, death, culling) over the planning horizon.

The cash flow summary details all sources of cash inflows (beginning cash, net cash income, off-farm income and interest earnings) and all cash outflows (principle payments, taxes, machinery down payments, etc.) to arrive at ending cash. The market value balance sheet is disaggregated sufficiently to show asset values and liabilities for each category of assets and liabilities on the farm (cash, crops in storage, land, machinery, livestock, etc.). Financial ratios summarizing rates of return, leverage, equity and debt for the farm are included with the balance sheet.

The income tax summary outputs begin with a detailed summary of the calculations for arriving at taxable income. These calculations are presented to resemble the federal income tax form known as Schedule F. The second tax summary is a one page table which details the steps used to calculate federal income tax, self-employment taxes, and Medicare taxes. For a sole proprietor the last tax summary page details the calculations for arriving at state income taxes. The model is capable of simulating a C-corporation, an S-corporation, a limited liability partnership, and a regular partnership. When one of the multiple owner organization forms is selected by the analyst, the model develops a summary of the additional tax calculations required for the particular organization and the cash requirements necessary to pay the other partners or share holders.

The final section of output for a deterministic analysis is a summary of the crop and livestock enterprises. In the crops section, the model prints a summary of annual acres, prices, yields, production, market receipts, government payments, production costs, and quantity fed to livestock for each crop. Each livestock enterprise has a summary table showing the herd dynamics (number born, died, culled, sold, replaced, etc.), prices received and paid, variable costs, receipts, feed requirements, and costs for purchased feeds.

When the model is run using stochastic crop yields, livestock production, and prices, the output is presented in terms of summary statistics. Each of the 4,000-plus output variables are summarized in terms of their means, standard deviations, coefficients of variation, minimums, and maximums. Additionally, the results are summarized in terms of probabilities, such as the probability of remaining solvent (surviving), probability of an economic success (return to equity greater than discount rate), and probability of losing more than X percent of real net worth. Annual probabilities of experiencing a cash flow deficit, of having to refinance equity to cover deficits, and of losing real net worth are also calculated. In total more than 200 probabilities summarizing the risk faced by the business are calculated.

The net present value (NPV) for each run (or scenario) has 100 values representing the 100 iterations of a stochastic analysis. These values when sorted from low to high constitute an empirical probability distribution for NPV. The NPV distributions for alternative scenarios can be compared using stochastic dominance with respect to a function to project producer preference among the scenarios analyzed.

During the 2002 Farm Bill devate, FLIPSIM was used to analyze about two dozen policy alternatives considered by the House Ag Committee. A side-by-side analysis of the House bill and the Senate bill for each of AFPC's representative farms was presented to the Ag Committees prior to the Conference Committee meeting. The House/Senate bill analysis was presented using cummulative probability distributions to show the safety net (risk reducing) impacts of the two bills on net incomes for representative farms. After the farm bill was passed, an analysis of alternative payment limitation provision's impacts on representative crop farms were presented to the Payment Limitation Commission.

Technology assessment studies with FLIPSIM have focused on quantifying the probable impacts on adopters and non-adopters. Technologies analyzed for livestock operations in the United States were bST and pST. For farms in Africa, the model was used to analyze payoffs to an infection/treatment method for controlling East Coast Fever in cattle. The benefits to adopting LEPA were analyzed for irrigated farms in Texas, as well as benefits of cotton, sorghum, and wheat variety improvements.

Farm management analyses which used FLIPSIM to quantify the firm level costs/benefits of alternative practices or decisions are quite numerous. Those of particular interest are:

• Multi-peril crop insurance,
• Revenue insurance,
• Alternative debt levels,
• Off farm income and family living impacts on farms,
• Farm growth strategies under alternative farm policies,
• Machinery purchase vs. leasing and alternative replacement strategies,
• Alternative federal and state income tax provisions,
• Drought effects on livestock producers,
• Increased price variability due to eliminating farm programs,
• Alternative farm program participation strategies
• Cash lease options of share vs. cash lease,
• Alternative farming practices to comply with conservation compliance,
• Best management practices for dealing with EPA restrictions,
• Green payment calculations for alternative farming practices.
• Payment limitations

²Not depicted in Figure 4 is the calculation for property taxes. Annual real estate taxes are calculated as the product of the market value of real estate in t-1 and a fixed property tax rate expressed as a dollar of tax per dollar of market value. Also not depicted in Figure 4 are the dairy and livestock enterprises.