• Adams, R. M., K. J. Bryant, and B. A. McCarl, 1995: Value of improved long range weather information. Contemp. Econ. Policy,13, 10–19.

  • Australian Bureau of Meteorology, 1991: Seasonal climate outlook No. 28—September to November 1991. National Climate Centre, Melbourne, Australia, 18 pp. [Available from Bureau of Meteorology, G. P. O. Box 1289K, Melbourne, VIC 3001, Australia.].

  • Global Climate Observing System, 1995: Socio-economic value of climate forecasts: Literature review and recommendations. GCOS-12 WMO/TD-674, World Meteorological Organization, Geneva, Switzerland, 34 pp.

  • Greenberg, J., 1976: Economic benefits of improved meteorological forecasts. Weather Forecasting and Weather Forecasts: Models, Systems and Users, Vol. 2, National Center for Atmospheric Research, 608–629.

  • Hill, S. J. H., J. W. Mjelde, W. Rosenthal, and P. J. Lamb, 1999: The potential impacts of the use of Southern Oscillation information on the Texas aggregate sorghum production. J. Climate, 12, 519–530.

  • Hilton, R. W., 1981: The determinants of information value: Synthesizing some general results. Manage. Sci.,27, 57–64.

  • Hollinger, S. E., 1988: Value of weather and climate information: Modeling weather management interactions of maize. Illinois State Water Survey Misc. Publ. 101, Champaign, IL, 30 pp. [Available from The Illinois State Water Survey, University of Illinois, 2204 Griffith Dr., Champaign, IL 61820.].

  • Hollinger, S. E., and R. C. Hoeft, 1986: Influence of weather on year-to-year corn response to ammonia fertilization. Agron. J.,78, 818–823.

  • Johnson, S. R., and M. T. Holt, 1986: The value of climate information. Policy Aspects of Climate Forecasting, R. Krasnow, Ed., Resources For the Future, 53–78.

  • Lave, L. B., 1963: The value of better weather information to the raisin industry. Econometrica,31, 151–164.

  • Livezey, R. E., 1990: Variability of skill of long-range forecasts and implications for their use and value. Bull. Amer. Meteor. Soc.,71, 300–309.

  • Mjelde, J. W., and M. J. Cochran, 1988: Obtaining lower and upper bounds on the value of seasonal climate forecasts as a function of risk preferences. West. J. Agric. Econ.,13, 184–192.

  • Mjelde, J. W., and S. E. Hollinger, 1989: Climate indices for application in empirical crop production studies. Agric. Syst.,30, 1–14.

  • Mjelde, J. W., S. T. Sonka, B. L. Dixon, and P. J. Lamb, 1988: Valuing forecast characteristics of a dynamic agricultural production system. Amer. J. Agric. Econ.,70, 674–684.

  • Mjelde, J. W., S. T. Sonka, and D. S. Peel, 1989: The socioeconomic value of climate and weather forecasting: A review. Midwestern Climate Center Research Rep. 89-01, Champaign, IL, 21 pp. [Available from the Midwestern Climate Center, University of Illinois, 2204 Griffith Dr., Champaign, IL 61820.].

  • Mjelde, J. W., D. S. Peel, S. T. Sonka, and P. J. Lamb, 1993: Characteristics of climate forecast quality: Implications for economic value to midwestern corn producers. J. Climate,6, 2175–2187.

  • Nicholls, J. M., 1996: Economic and social benefits of climatology information and services: A review of existing assessments. World Meteorological Organization, Geneva, Switzerland, WMO/TD-780, 38 pp.

  • Paltridge, G. W., 1985: The value of climate forecasting. Geophys. Surv.,7, 273–290.

  • Penson, J. B., Jr., and B. L. Gardner, 1988: Implications of macroeconomic outcomes for agriculture. Amer. J. Agric. Econ.,70, 1013–1022.

  • Penson, J. B., and C. R. Taylor, 1992: United States agriculture and the general economy: Modeling their interface. Agric. Syst.,39, 33–66.

  • Penson, J. B., D. A. Lins, and G. D. Irwin, 1971: Flow of funds social accounts for the farm sector. Amer. J. Agric. Econ.,53, 1–7.

  • Reetz, H. F., Jr., 1976: CORN CROPS! A physiology-based simulation of the corn crop. Ph.D. dissertation, Purdue University, West Lafayette, IN, 159 pp. [Available from Life Sciences Library at Purdue University, Purdue University, West Lafayette, IN 47907.].

  • Romain, R. F. J., J. B. Penson Jr., and R. E. Lambert, 1987: Capacity depreciation, implicit rental price, and investment demand for farm tractors in Canada. Can. J. Agric. Econ.,35, 373–378.

  • U.S.D.A., 1996a: Agricultural Statistics. National Agricultural Statistics Service, Washington, D.C., 511 pp. [Available from the U.S.D.A., 14th St. and Independence Ave. SW, Washington, DC 20250.].

  • U.S.D.A., 1996b: Grain: World markets and trade. Foreign Agricultural Service, Circular Series FG 3-96, Washington, D.C., 56 pp. [Available from the U.S.D.A., 14th St. and Independence Ave. SW, Washington, DC 20250.].

  • U.S. Dept. of Commerce, 1996: Statistical Abstract of the United States. 116 ed. Economics and Statistics Administration, Bureau of Census, 1022 pp.

  • Weston, J. F., and E. F. Brigham, 1990: Essentials of Managerial Finance. 9th ed. The Dryden Press, 931 pp.

  • Wilks, D. S., 1997: Forecast value: Prescriptive decision studies. Economic Value of Weather and Climate Forecast, R. Katz and A. Murphy, Eds., Cambridge University Press, 109–149.

  • View in gallery
    Fig. 1.

    Schematic of the relationships in the AG-GEM model.

  • View in gallery
    Fig. 2.

    Differences (with forecast − without forecast) in the Corn Belt region for debt burden ratio for the five timing scenarios.

  • View in gallery
    Fig. 3.

    Differences (with forecast − without forecast) in the Corn Belt region for present value of ending wealth for the five timing scenarios.

  • View in gallery
    Fig. 4.

    Differences (with forecast − without forecast) in the Corn Belt region for return on equity ratio for the five timing scenarios.

  • View in gallery
    Fig. 5.

    Differences (with forecast − without forecast) in national present value of producer surplus for the five timing scenarios.

  • View in gallery
    Fig. 6.

    Differences (with forecast − without forecast) in national present value of consumer surplus for the five timing scenarios.

  • View in gallery
    Fig. 7.

    Differences (with forecast − without forecast) in national present value of consumer plus producer surplus for the five timing scenarios.

  • View in gallery
    Fig. 8.

    Differences (with forecast − without forecast) in the Corn Belt region for present value of machinery and motor vehicles for the five timing scenarios.

All Time Past Year Past 30 Days
Abstract Views 0 0 0
Full Text Views 180 36 12
PDF Downloads 39 15 0

Dynamic Aspects of the Impact of the Use of Perfect Climate Forecasts in the Corn Belt Region

James W. MjeldeDepartment of Agricultural Economics, Texas A&M University, College Station, Texas

Search for other papers by James W. Mjelde in
Current site
Google Scholar
PubMed
Close
,
John B. Penson Jr.Department of Agricultural Economics, Texas A&M University, College Station, Texas

Search for other papers by John B. Penson Jr. in
Current site
Google Scholar
PubMed
Close
, and
Clair J. NixonDepartment of Accounting, Texas A&M University, College Station, Texas

Search for other papers by Clair J. Nixon in
Current site
Google Scholar
PubMed
Close
Full access

Abstract

A general equilibrium model is linked to a decision model to determine the impact of perfect growing season forecasts for corn produced in the Corn Belt region over a 10-yr period. Five different timing scenarios are examined to determine the effect of different orderings in the occurrence of good and bad crop years over this period. The use of the climate forecasts is shown to have both positive and negative financial and economic effects depending on the specific year within any given scenario. The expected present value of changes in net surplus (consumer plus producer surplus) varied from $1.270 to $2.917 billion from the use of the perfect forecasts over different 10-yr planning horizons. Consumers are the clear winners (positive values) and producers are the losers (negative values) over the entire horizon.

Corresponding author address: Prof. James W. Mjelde, Department of Agricultural Economics, Texas A&M University, Blocker Building, College Station, TX 77843-2124.

j-mjelde@tamu.edu

Abstract

A general equilibrium model is linked to a decision model to determine the impact of perfect growing season forecasts for corn produced in the Corn Belt region over a 10-yr period. Five different timing scenarios are examined to determine the effect of different orderings in the occurrence of good and bad crop years over this period. The use of the climate forecasts is shown to have both positive and negative financial and economic effects depending on the specific year within any given scenario. The expected present value of changes in net surplus (consumer plus producer surplus) varied from $1.270 to $2.917 billion from the use of the perfect forecasts over different 10-yr planning horizons. Consumers are the clear winners (positive values) and producers are the losers (negative values) over the entire horizon.

Corresponding author address: Prof. James W. Mjelde, Department of Agricultural Economics, Texas A&M University, Blocker Building, College Station, TX 77843-2124.

j-mjelde@tamu.edu

Introduction

The effects of the droughts of 1996 and 1998, the 1997–98 El Niño event, and the 1998–99 La Niña event on fluctuations in commodity prices reinforce the dependence of agriculture on weather/climatic conditions. Recent advances in the sciences of meteorology and oceanography indicate the ability to forecast climate (weather occurring over a period of approximately 2 weeks or longer) conditions is becoming a reality (Livezey 1990; Australian Bureau of Meteorology 1991). Along with these advances has been a proliferation of studies examining the effect of climate fluctuations and the use of climate forecasts on society [for reviews of this literature see Wilks (1997); Nicholls (1996); Global Climate Observing System (1995); or Mjelde et al. (1989)]. These studies suggest agriculture, at the farm level, is able to adjust to take advantage of improved forecasts. Few studies, however, have addressed this issue from a sector-level perspective, and no studies have examined linkages to the broader economy. Both Global Climate Observing System (1995) and Mjelde et al. (1989) state the need to conduct such analyses to further our understanding of the impact of the use of improved climate forecasts in agriculture.

Paltridge (1985), using a highly simplified example, illustrates that improved climate forecasts may be detrimental to the producer. Lave (1963), using an econometrics approach, showed improved climate forecasts may reduce the raisin industry’s net returns because of an increase in output and a corresponding larger decrease in price. On the other hand, Greenburg (1976) suggests all benefits from improved climate forecasts accrue to the producer. Johnson and Holt (1986) present a rational expectation framework in a theoretical setting. Their framework is, however, difficult to use in applied settings because of data and aggregation limitations.

One sector-level study that addressed both consumer and producer welfare is Adams et al. (1995). They address the value of improved Southern Oscillation forecasts in a long-run equilibrium framework. Their results indicate the annual economic value (consumer plus producer surplus) of Southern Oscillation forecasts for the southeastern United States is between $96 and $130 million annually (1990 dollars). Similarly, Hill et al. (1999) addressed the aggregate impact of the use of Southern Oscillation forecasts on sorghum production in Texas. Their results suggest the use of climate information may increase or decrease expected Texas sorghum supply relative to the expected supply generated without the use of the information.

Missing from these studies is any link to sectors other than production agriculture. Further, these previous studies have ignored the impact of timing of the climate conditions; that is, they have obtained the expected annual value of climate forecasts. Timing was not an issue. The effects of these two components, linkages to other sectors and timing, have not received any attention in the literature. Through a highly simplified example, the objective of the current study is to illustrate the need to address multisector and timing issues if a complete understanding of the impact of the use of climate forecasts is to be obtained. To accomplish this objective, this study examines the impact of perfect climate forecasts within the context of an agricultural general equilibrium model, AG-GEM, which captures the direct and indirect effects over time. This is the first study 1) to take the impacts of improved climate forecasts beyond the agriculture sector to capture some broader impacts on the economy and 2) explicitly to consider the time paths of the occurrence of climate conditions. An important component in AG-GEM is its ability to simulate the U.S. economy with special emphasis on agriculture over an extended period of time. Here, expected net present value for key performance variables is obtained rather than focusing upon expected economic value at a specific point in time.

The unrealistic case of perfect forecasts is used to illustrate some important aspects of the impact of the use of improved climate forecasts that have not been previously discussed. Using perfect forecasts simplifies the modeling and discussion without changing the general inferences drawn from the results. It is hoped this study will stimulate additional research and discussion on the importance of linkages within the economy and the timing of the occurrence of events. As noted, results from previous studies have presented a single expected value. Like all averages, such a value, while useful, extrapolates from many issues as is illustrated by the results.

Methodology

To examine the impact of improved climate forecasts at the sector level and beyond, two economic models are used. The first is a field-level decision model, which determines the impact of improved forecasts on yields, input usage, and costs. These changes are then incorporated into the AG-GEM econometrics simulation model. Before each of these models is discussed, the climate forecast scenario is presented.

Climate forecast scenario

Only the extreme case of perfect climate forecasts for corn production in the Corn Belt region is examined. This partial equilibrium approach is used for several reasons. Corn, one of the most important U.S. crops grown and exported, accounted for approximately 17% of the cash receipts from farm marketing of crops in the United States in the period 1992–94 (U.S. Dept. of Commerce 1996). In addition, U.S. exports accounted for approximately 70% of the world corn trade during the 1992/93–1994/95 period (U.S.D.A. 1996b). The Corn Belt region (Illinois, Iowa, Indiana, and Ohio) produces approximately 50% of the U.S. corn production (U.S.D.A. 1996a). Yield and cost changes for corn caused by the use of forecasts in this region should affect the agriculture sector and beyond. Further, only changes in the Corn Belt region are modeled because previous literature indicates the ability to forecast climate conditions may vary by region (Livezey 1990). Another reason for modeling only changes associated with corn in the Corn Belt region is the availability of a corn production model designed to examine input changes associated with improved yield changes based on improved climate forecasts for Illinois (Mjelde et al. 1988).

Given the multitude of potential improvements in climate forecasts, only the extreme case of perfect forecasts is examined here. Such an analysis, although admittedly overly optimistic, simplifies the modeling and presentation of the results, while maintaining the potential impacts of improved climate forecasts. This simplification occurs because of the timing of the years, which is a major issue in this study. With 10 years, and seven forecasts necessary for each year (see below), the potential number of combinations is enormous. While examining such a large number of combinations for an imperfect forecast is more realistic, it does not change the inferences drawn from the results. Further, given the linkages and adjustments available, if changes resulting from the use of improved climate forecasts are present, they should appear under the extreme case of perfect information.

Farm-level model

A risk-neutral, dynamic programming model of corn production in Illinois is used to determine how producers would change input usage in response to improved climate forecasts. This model is an updated version of the model presented in Mjelde et al. (1988). The model is capable of examining “. . . how attributes of the decision maker’s environment and characteristics of the forecast, in addition to accuracy, affect the economic value of climate forecasts” (Mjelde et al. 1988 p. 674). The model has been accepted by the agricultural (Mjelde et al. 1988; Mjelde and Cochran 1988) and meteorological (Mjelde et al. 1993) professions.

Of key importance is the sequential nature of corn production. Within the model, the production season is divided into eight stages corresponding to the timing of major input production decisions (Table 1). Decisions included are fertilizer application (amount and timing), planting date, planting density, hybrid selection, and harvest date. Production stages are previous fall, early spring, late spring, early summer, midsummer, late summer, early harvest, and late harvest. A modified version of a corn-growth simulation model developed by Reetz is used to provide the data for the relationship between climatic conditions and corn yield (Reetz 1976; Hollinger and Hoeft 1986). In general, the simulated yields closely resembled actual yields from Illinois (Hollinger 1988).

Climate conditions and forecasts for the early spring to early harvest stages are quantified using a climate index based on plant growth (Mjelde and Hollinger 1989). This index incorporates daily precipitation, temperature, and solar radiation into a relative growth rate for corn. Between previous fall and early spring, winter precipitation is forecasted, whereas between early and late harvest, the forecast is based on corn drydown potential. Corn drying potential is dependent on temperature and precipitation. As such this study does not focus on the meteorological process that might give rise to a forecast, but rather on what are some of the potential implications of the use of climate forecasts.

In the model, climate conditions enter in three different relationships depending on the production stage. Between the previous fall and early spring, winter precipitation along with the amount of fall applied nitrogen determines the level of leaching and denitrification that occurs. Applying nitrogen in the fall allows for additional field time for cultivation and planting operations in the spring, but the producer runs the risk of losing some of the fall applied nitrogen. Loss of nitrogen is a cost to the producer. During the early and late spring, field time is limited based on the climate conditions occurring. The model examines the tradeoff between the potential loss of nitrogen and the increased availability of field time (not necessary to apply nitrogen in the spring) for operations other than nitrogen application. The relative growth index for growing season conditions enters the model through an estimated dynamic production function. The levels and interactions between climate conditions and input usage helps determine corn yield. At maturity, corn must be dried either artificially or in the field for safe storage. Harvesting at a higher corn moisture level can increase yields, but the producer must incur higher drying costs. Letting the corn dry in the field lowers drying costs, but increases the potential for field losses because of inclement climate conditions. In the model, corn drying potential based on temperature and precipitation is used to help to determine if the corn crop should be harvested at early harvest or late harvest. It is the interactions between climate conditions and input usage that make the use of climate forecasts potentially valuable to producers. The producers may be able to lower costs, increase yields, or both if the climate conditions are known. For all periods, the three indices (winter precipitation, relative growth, and drydown) are placed into either poor, average, or good growing conditions categories.

Because AG-GEM is limited to simulating 10 yr, only the last 10 of the 14 yr presented in Mjelde et al. (1988) provide the historical base for this study. The decision model is used to determine the percent change of yields and costs per acre from the historical base caused by the use of climate information (Table 2). The use of these percentage differences in AG-GEM is discussed in the next section. Percentage differences are used based on previous findings that modeling percentage differences rather than absolutes provides better aggregate estimates of changes in yields.

The following procedure is used to obtain the percentage differences in yields and variable costs. First, the decision model is solved using only historical probabilities of the various climate conditions based on the 10-yr base. The optimization procedure provides the optimal decision for all possible states (combinations of previous decisions and climate conditions) in the current decision period. For example, the model provides an optimal decision set in early spring for all possible fall applied nitrogen levels and winter precipitation levels. Similarly, an optimal decision is given for all previous conditions and management input levels for late spring through late harvest. At any decision point, the optimal decisions are based on the current state of the process and expected climate conditions. Expected climate conditions are based on the historical probability of the three possible climate condition categories. Using these optimal decisions by production period, simulations for climate conditions for each of the 10 yr can be conducted by inputting the actual climate conditions occurring in a particular year. Associated yields and variable costs for each year are obtained. Percentage differences for each year are then calculated using the mean yield and variable costs.

For the perfect forecast scenario, the decision model is solved for the climate conditions associated with each year. Yield and variable costs are obtained directly from the decision model. Percentage differences in yields and costs are obtained using the historical mean yield and costs. This approach is consistent with previous studies that use a decision-theory approach to value information (Hilton 1981).

Percentage differences presented in Table 2 indicate the use of climate forecast information has different impacts on yields and costs depending on the year. In five of the years, the optimal historical decisions are identical to the optimal perfect forecast decisions (shown by no changes in the percentage differences in yields and costs). For three of the years, the use of the climate forecasts decreased costs along with a corresponding increase in yields relative to the historical percentage changes in costs and yields. Another year shows a decrease in costs and yields. In one year an increase in cost is indicated with a corresponding increase in yield. These results are consistent with Hill et al. (1999) findings that the use of climate information may increase or decrease yields, but the use of climate forecasts allows the decision maker to more efficiently use inputs.

The percentage changes in yields and costs are caused by changes in input usage. One may wonder why the use of a perfect forecast would decrease yields. Consider year 2, a poor year in terms of overall growing conditions (Table 2). Using decisions based on historical knowledge, yields are 9% less than the historical average. The use of a perfect forecast decreases yields 13% over the historical average, approximately a 4% decrease in yields that is caused by the use of the forecasts. The reason can be found in the cost column. With perfect knowledge, producers will decrease their input usage below the level used when only historical knowledge is used (14% decrease in expected costs calculated as the 6% increase in costs associated with historically based decision rules and the 8% decrease associated with the perfect knowledge decisions). The decrease in input usage decreases yields but increases overall net returns because the decreased input usage also decreases costs. Costs decrease more than the decrease in revenues caused by the lower yields (14% versus 4%).

In determining the optimal inputs and associated yields and costs, an expected price of $2.40 bu−1 is assumed. This price is the Olympic average (highest and lowest prices are removed from the calculation) of yearly corn prices in the Corn Belt region of the United States for the years 1979–93. Corn price and input costs are assumed to be fixed in the model. Variable costs include expenditures for nitrogen and seeds, application charges, harvest costs, and drying charges. In addition, an interest charge is assessed on expenditures. Changes in nitrogen application rates provided the largest change in input usage and variable costs.

To examine the impact of the timing of good and bad crop years, yield and cost changes are utilized by AG-GEM in five unique sequences: 1) actual occurrence, 2) reverse order, 3) worst years first, 4) best years first, and 5) randomly drawn. The actual occurrence scenario utilizes the annual data in the order they actually occurred, that is, they are utilized in the order 1–10. Thus, the first year of the simulation is year 1 in Table 2, the second year is year 2 in the table, and so on. The reverse order scenario assumes the years occur in the opposite order given in Table 2. The worst years first scenario assumes the years are ordered from the worst year in terms of historical yield to the highest yield (order of the years is 10, 9, 2, 5, 8, 4, 1, 6, 7, and 3). The best years first scenario assumes the years occur from the highest yield to the lowest yield, creating a scenario just the opposite of the worst year scenario. Last, the order of the years was randomly drawn giving an order of 1, 9, 4, 10, 7, 2, 3, 8, 6, and 5.

AG-GEM

As previously noted, the results from the decision model are utilized by AG-GEM, a model that links the agricultural sector to the broader U.S. macroeconomy. As such, this study is the first attempt to address the impact of the use of climate forecasts beyond the agricultural sector. An earlier version of this model is described in Penson and Taylor (1992). AG-GEM, a large-scale macroeconomic simulation model of the U.S. economy, has been continuously developed over the past 25 or more years. As shown in Fig. 1, this model places major emphasis on agriculture and its linkages to related sectors that 1) supply farm inputs; 2) purchase farm production for further processing; 3) provide short-, intermediate-, and long-term loans to farmers and ranchers; and 4) provide hired labor services to farms and ranches. The model accounts for substitution across crops and the feedback relationships between crops and livestock. Fourteen major crops and crop by-products and 14 major livestock and livestock by-products are modeled annually on an individual basis for each of 10 production regions. Changes in the agricultural sector can be estimated for their effect not only on consumer and producer surplus but also on variables such as financial ratios or term debt. The model also calculates the direct and indirect effects changes in monetary and fiscal policies have upon agriculture. AG-GEM simulates the linkages and national impacts on an annual basis for 10 years into the future.

The solution of the AG-GEM model rests on assumptions made with respect to future trends in such exogenous variables as the rate of growth in the money supply, income, and capital gains tax rates; government spending; farm policy affecting direct payments to farmers; and trade legislation affecting tariffs and quotas. Given these assumptions, AG-GEM initially projects future trends in interest rates, inflation rates, unemployment rates, real gross domestic product growth rates, and foreign exchange rates. AG-GEM also projects the unit cost of materials used in manufacturing farm inputs. These unit costs are used, for example, in explaining variations in the cost of production for such inputs as fertilizer and chemicals, as well as fuel and energy. Trends in each of these variables influence the operation and expansion of farms and ranches at the regional level, which in turn affects the supply of crop and livestock products.

AG-GEM projects the food, feed, export, stock, and other demands for individual crop and livestock prices. In doing so, the model accounts for the simultaneity between crops and livestock in solving for market equilibrium commodity prices. More specifically, the AG-GEM model provides the following output for the agricultural sector:

  • commodity prices for specific crops and livestock;

  • supply and use for each commodity;

  • national and regional pro forma financial statements (balance sheets, income statements, and cash flow statements);

  • capital expenditures for categories of farm machinery and motor vehicles;

  • variable input use for fertilizer, chemicals, fuel, repairs, and labor;

  • net returns by commodity in each of the 10 production regions; and

  • trends in farm land values.

Thousands of equations (statistically estimated equations and identities) based on macro- and microeconomic theory compose AG-GEM. As with all large simulation models, it is impossible to provide all the intricate details of the model. AG-GEM has been extensively used in public policy research and outlook forums. The model was used, for example, to analyze the effects of the Freedom to Farm Act, which ultimately became known as the FAIR Act when signed into law in 1996. This included studies for the Farm Credit Council and Farm Credit Administration, the trade and regulatory components of the $60 billion Farm Credit System, a major lender to farmers and ranchers. This model also was used to demonstrate the effects of alternative policy options to staff of the U.S. Congress during the debate on the FAIR Act. Many of the components of AG-GEM have been documented in leading national and international journals (Penson et al. 1971; Romain et al. 1987; Penson and Gardner 1988).

For purposes of this study, cost and yield changes associated with the 10 yr used in the field-level decision model are used to simulate the years 1996–2005 in AG-GEM. It should be noted the financial conditions necessary for AG-GEM are projections to the year 2005, but the climate conditions are as given in Table 2. No changes in the financial conditions are simulated, other than those caused by changes in yields and costs. The results are for the potential impact of improved climate forecasts and should not be viewed as actual forecasts for the years 1996–2005. Further, the forecasts are 10 1-yr forecasts received annually, and not 1 10-yr forecast.

Cost and yield modifications are made to AG-GEM to account for the potential impact of improved climate forecasts. Variable production costs in the projected Corn Belt regional enterprise production budgets in AG-GEM are modified by the percentages given in Table 2 for the various scenarios. Two different modifications are necessary for yields, because of the regional and national modeling in AG-GEM. First, Corn Belt regional yields are modified from the baseline trends in AG-GEM according to the percentages in Table 2. Second, national yields are modified. National yields are adjusted by multiplying the yield baseline by (1 + 0.5x), where x is the yield percentage differences given in Table 2. A factor of 0.5 is used because historically approximately 50% of the U.S. corn production occurs in the Corn Belt region. All modifications occur for both the historical and perfect forecast scenarios under the four different scenarios.

Because one of the goals of this study is to determine the dynamic economic impact from the use of climate forecasts, discounting becomes an important issue. Annual discount rates used reflect projected future trends in the prime interest rate at commercial banks as determined by AG-GEM. The annual rates vary from 9.4% to 10.5%.

Results

Five general categories of results are presented: 1) financial stress, 2) wealth effects, 3) profitability, 4) societal economic welfare impacts, and 5) acquisition of farm inputs. Because this study addresses the impact of the use of climate forecasts, differences in these categories between the values when the forecasts are used and without the forecasts are presented. Debt burden is presented as the measure of financial stress. Present value of ending annual wealth is used as an indicator of changes in wealth position from the use of improved climate forecasts. The rate of return on equity is used to illustrate the impact of changing market prices and production costs on profits. Producer and consumer surplus are used to illustrate the overall economic impact to society. Purchases of machinery and motor vehicles are used to illustrate impacts on the input sector. All five categories except societal welfare impacts are presented for the Corn Belt region. Welfare impacts are calculated at the national level, because AG-GEM clears markets at the national level.

Financial stress

The debt burden ratio measures outstanding term debt in relation to net cash farm income. This ratio indicates the number of years it would take to retire outstanding term debt if net income remained at the current level. The greater the debt burden ratio, the more susceptible producers are to financial risk if adverse trends in interest rates or net income occur. Therefore, the lower the value of this ratio, the lower the exposure to financial risk. Negative differences between the “with” and “without” forecast simulations indicate the use of the forecasts has lessened the financial stress experienced by producers within the Corn Belt region. A negative difference indicates debt relative to net cash farm income is higher in the without forecast simulation, whereas a positive difference indicates the opposite has occurred.

Differences in the debt burden ratio between the with and without forecast simulations are presented in Fig. 2 for the five scenarios. The differences in the time path for each scenario indicate the degree of volatility of term debt in relation to net cash farm income because interest rates exhibit little change among the scenarios. These results appear reasonable given the relationship between the occurrence of the poor performance years and the borrowing needs of producers. In the reverse and worst year scenarios, the worst growing year occurs first, causing early differences in this ratio to be negative. In the actual and best year scenarios, an opposite pattern is present, namely, a large negative difference in the last year. Further, there are small differences between the actual and best year scenarios. The randomly drawn scenario shows the impact of the worst year in the middle of the planning horizon.

As a specific example, consider the first year of the worst year scenario. In this year the with forecast debt burden ratio is 6.9, whereas the without forecast ratio is 11.0, for a difference of −4.1. This difference suggests the use of the climate forecast decreased the number of years required to retire debt by over four years if net cash income remained at the current level. An even more dramatic difference is shown in the last year of the actual occurrence scenario. In this situation, the debt burden ratios are 16.6 and 148.3 for a difference of −131.7. The yields and costs associated with these two scenarios are identical (year 10 in Table 1). The worst year scenario difference in debt burden for the last year in the planning horizon is −0.002, compared with −131.7 for the actual occurrence scenario. In both scenarios, because it is the end of the planning horizon, all 10 yr have been experienced. These examples show the importance of the timing of the years in determining financial stress.

Wealth effects

The present value of ending annual wealth is used as an indicator of how wealth is affected by the use of climate forecasts. A positive difference in ending wealth between the with and without simulations indicates the use of the climate forecasts has increased wealth relative to the without simulation.

Shown in Fig. 3 are differences in the present value of ending wealth. Differences of almost $2.0 billion are evident in some years of the planning horizon. The largest changes in the differences in wealth occur the year following the year with the smallest yields. This year is year 2 in the reversed and worst year scenarios, and year 6 in the randomly drawn scenario. If this pattern holds true, for the actual and best year scenarios, the impact of the low yields may not be felt until after the simulation ends (year 11). Of the 50 yr in Fig. 3 (10 years per scenario and 5 scenarios), 29 show a decrease in wealth caused by the use of forecasts, 4 show no change, and 17 show an increase in wealth. Only the best year scenario has an equal number of years with increases and decreases in wealth. In all other scenarios, more years show a decrease than an increase in the present value of ending wealth.

In the three scenarios, actual, reversed, and best year, the present value of annual ending wealth differences are positive at the end of the time horizon. For the worst year and randomly drawn scenarios, the difference in present value of ending annual wealth is negative at the end of the planning horizon. Differences in wealth between scenarios range from $−77 million to over $428 million at the end of the horizon. These differences again highlight the importance of considering the order of occurrence of climate events in determining the impact of the use of climate forecasts.

Profitability

Profitability ratios show the combined effects of decisions on debt management, asset management, input usage, etc. (Weston and Brigham 1990). Presented in Fig. 4 is return on equity (ROE) or net farm cash income divided by farm equity. This ratio indicates the returns to producer capital. Positive differences in ROE between the with and without forecast simulations indicate the use of climate forecasts increases profitability.

Poor performances in the early years (after the first year) of the planning horizon had a relatively large impact on the ROE in the reversed and worst year scenarios. Actual and best year scenarios showed greater differences near the end of the planning horizon. In the randomly drawn scenario, the largest differences occur in the middle of the horizon. As shown in Fig. 4, the use of climate forecasts can increase or decrease profitability depending on the specific year within each scenario. Under all scenarios, the use of climate forecasts increased profitability in the year with the smallest yield (year 10 in Table 2). The lowest yields occurs in year 10 in the actual and best year scenarios, year 1 in the reverse and worst year scenarios, and year 5 in the randomly drawn scenario.

Societal economic welfare impacts

To capture the overall societal economic impact from the use of the climate forecasts, three economic welfare measures are presented: producer surplus, consumer surplus, and net welfare. Consumer surplus is measured as the total value of the use of a product minus its value in exchange. Producer surplus represents the net gain to producers. Net welfare is the sum of producer plus consumer surplus. Producer and consumer surplus are measured at the national level and include all crop markets contained in AG-GEM. Positive differences between the with and without forecast simulations indicate an increase in the surplus measure as a result of the use of the climate forecast.

Producer surplus differences are dramatic for the reversed and worst year scenarios in the early years of the planning horizon and for actual and best year scenarios in the later years (Fig. 5). For the randomly drawn scenario, losses in producer surplus are not as dramatic for the worst growing year of the planning horizon. In the remaining years, producer surplus differences are not as large but generally are negative. In any given year, the difference could be positive. The effect of increased yields through input changes is offset by market price adjustments. These results suggest producers, as a whole, are better off following input usage without the forecasts than utilizing the climate information.

Measures of the overall economic impact on society are illustrated in Figs. 6 (consumer surplus) and 7 (net welfare). As illustrated in Fig. 6, consumers, as a whole, generally benefit from the use of improved climate forecasting. The greatest consumer surplus differences occur in the years that had the largest decrease in producer surplus. The total change in consumer plus producer surplus, as shown in Fig. 6, exhibits a positive effect from the use of climate forecasts. For the most part, the differences are positive, suggesting society is a net winner when producers have better climate information available. Still, depending on the sequence and the year in question, the without forecast simulation may provide greater net benefits to society for a particular year than the with simulation.

The present values over the 10-yr time horizon for producer, consumer, and net surplus changes are given in Table 3. Changes in producer surplus range from$−7238 to $−4440 million. Using $−4440 million as the base, this is a range of 63% in absolute value. Consumer surplus changes ranges from $6328 to $9980 million, a range of 58% (base of $6328). Net surplus changes vary more than either producer or consumer surplus, ranging from $1270 to $2917 million or 130%. For any given year, a positive producer surplus can be associated with a positive or negative consumer surplus and vice versa. In all five scenarios, at least three years had a negative net surplus. These results indicate the use of climate forecasts overall increases society’s welfare, but in any given year society may experience a net surplus loss or gain. Such findings are not evident in the previous static analyses.

Acquisition of farm inputs

To illustrate the spinoff effects of the use of climate forecasts, the annual change in the present value of farm machinery and motor vehicles is presented. These changes provide an indication of activity in the farm input sector. Positive differences between the with and without forecast simulations indicate increased purchases of machinery and vehicles representing a positive spinoff effect for the input sector.

Presented in Fig. 8 is the present value of differences in farm machinery and motor vehicles value. In the actual and best year scenarios the differences are relatively small until the last year, whereas in the other three scenarios differences are much greater earlier in the planning horizon. The differences in all five scenarios are negative in at least 6 of the 10 yr, indicating decreases in input purchases. Such decreases have ramifications on input dealers and rural economies as producers spend less on machinery and motor vehicles.

Conclusions

The results demonstrate the impact of the timing of good and bad crop years on financial conditions at both the sector and economy level when considering the impact of the use of improved climate forecasts. For example, depending on the occurrence of the years, the present value of net consumer and producer surplus ranged over 130%. Results of the other financial factors and spinoff activities further reinforce the importance of the timing of the good and bad years.

Economic welfare results indicate that society generally benefits from the use of improved climate forecasts. Over the 10-yr horizon, regardless of scenario, producers experience a negative impact while consumers experience a positive impact. Consumer benefits outweigh the negative producer effect. The above does not hold, however, for individual years. Namely, for any given year, producer or consumer surplus could be positive or negative. These results indicate the impact of the use of climate forecasts will vary by year, reinforcing the importance of timing. Such inferences are not available in previous static analyses of the impact of the use of climate forecasts. Further, the use of improved climate forecasts will have impacts beyond producers and consumers. As shown, the use of climate forecasts will impact factors such as the purchase of machinery. Such changes will, in turn, impact the general economy and the rural economy in particular. These impacts, however, depend on the timing of the good and bad crop years. Such impacts have not been previously discussed in the literature valuing climate forecasts.

This study is not designed to provide a definitive answer to the impact of the use of improved climate forecasts, but rather to illustrate the need to address multisector and timing issues if a complete understanding of the impact of the use of climate forecasts is to be obtained. An obvious limitation is examining only the use of perfect climate forecasts in corn production in the Corn Belt region. The goal of this study is to examine some dynamic impacts of the use of climate forecasts, and, as such, simplifications are required. Despite this limitation, this study does provide new insights into the impact of the use of improved forecasts that are not present in the current literature. Further, the study opens up many new avenues of additional research needed at the sector level and beyond. These research questions include 1) assessing the role of forecast quality in determining when impacts start appearing, 2) allowing for crop mix decisions dependent on the forecasts, 3) analyzing and quantifying the impact on rural economies, 4) determining the impact of including forecasts for more than one region, 5) quantifying the impact at the general economy level and global level, and 6) studying the effect of climate forecasts on price expectations. Addressing these and other issues will assist in better understanding of the effect of technology and science on the use of improved information by producers.

Acknowledgments

This research was partially supported by Department of Commerce National Oceanic and Atmospheric Administration Grant NA56GP0266.

REFERENCES

  • Adams, R. M., K. J. Bryant, and B. A. McCarl, 1995: Value of improved long range weather information. Contemp. Econ. Policy,13, 10–19.

  • Australian Bureau of Meteorology, 1991: Seasonal climate outlook No. 28—September to November 1991. National Climate Centre, Melbourne, Australia, 18 pp. [Available from Bureau of Meteorology, G. P. O. Box 1289K, Melbourne, VIC 3001, Australia.].

  • Global Climate Observing System, 1995: Socio-economic value of climate forecasts: Literature review and recommendations. GCOS-12 WMO/TD-674, World Meteorological Organization, Geneva, Switzerland, 34 pp.

  • Greenberg, J., 1976: Economic benefits of improved meteorological forecasts. Weather Forecasting and Weather Forecasts: Models, Systems and Users, Vol. 2, National Center for Atmospheric Research, 608–629.

  • Hill, S. J. H., J. W. Mjelde, W. Rosenthal, and P. J. Lamb, 1999: The potential impacts of the use of Southern Oscillation information on the Texas aggregate sorghum production. J. Climate, 12, 519–530.

  • Hilton, R. W., 1981: The determinants of information value: Synthesizing some general results. Manage. Sci.,27, 57–64.

  • Hollinger, S. E., 1988: Value of weather and climate information: Modeling weather management interactions of maize. Illinois State Water Survey Misc. Publ. 101, Champaign, IL, 30 pp. [Available from The Illinois State Water Survey, University of Illinois, 2204 Griffith Dr., Champaign, IL 61820.].

  • Hollinger, S. E., and R. C. Hoeft, 1986: Influence of weather on year-to-year corn response to ammonia fertilization. Agron. J.,78, 818–823.

  • Johnson, S. R., and M. T. Holt, 1986: The value of climate information. Policy Aspects of Climate Forecasting, R. Krasnow, Ed., Resources For the Future, 53–78.

  • Lave, L. B., 1963: The value of better weather information to the raisin industry. Econometrica,31, 151–164.

  • Livezey, R. E., 1990: Variability of skill of long-range forecasts and implications for their use and value. Bull. Amer. Meteor. Soc.,71, 300–309.

  • Mjelde, J. W., and M. J. Cochran, 1988: Obtaining lower and upper bounds on the value of seasonal climate forecasts as a function of risk preferences. West. J. Agric. Econ.,13, 184–192.

  • Mjelde, J. W., and S. E. Hollinger, 1989: Climate indices for application in empirical crop production studies. Agric. Syst.,30, 1–14.

  • Mjelde, J. W., S. T. Sonka, B. L. Dixon, and P. J. Lamb, 1988: Valuing forecast characteristics of a dynamic agricultural production system. Amer. J. Agric. Econ.,70, 674–684.

  • Mjelde, J. W., S. T. Sonka, and D. S. Peel, 1989: The socioeconomic value of climate and weather forecasting: A review. Midwestern Climate Center Research Rep. 89-01, Champaign, IL, 21 pp. [Available from the Midwestern Climate Center, University of Illinois, 2204 Griffith Dr., Champaign, IL 61820.].

  • Mjelde, J. W., D. S. Peel, S. T. Sonka, and P. J. Lamb, 1993: Characteristics of climate forecast quality: Implications for economic value to midwestern corn producers. J. Climate,6, 2175–2187.

  • Nicholls, J. M., 1996: Economic and social benefits of climatology information and services: A review of existing assessments. World Meteorological Organization, Geneva, Switzerland, WMO/TD-780, 38 pp.

  • Paltridge, G. W., 1985: The value of climate forecasting. Geophys. Surv.,7, 273–290.

  • Penson, J. B., Jr., and B. L. Gardner, 1988: Implications of macroeconomic outcomes for agriculture. Amer. J. Agric. Econ.,70, 1013–1022.

  • Penson, J. B., and C. R. Taylor, 1992: United States agriculture and the general economy: Modeling their interface. Agric. Syst.,39, 33–66.

  • Penson, J. B., D. A. Lins, and G. D. Irwin, 1971: Flow of funds social accounts for the farm sector. Amer. J. Agric. Econ.,53, 1–7.

  • Reetz, H. F., Jr., 1976: CORN CROPS! A physiology-based simulation of the corn crop. Ph.D. dissertation, Purdue University, West Lafayette, IN, 159 pp. [Available from Life Sciences Library at Purdue University, Purdue University, West Lafayette, IN 47907.].

  • Romain, R. F. J., J. B. Penson Jr., and R. E. Lambert, 1987: Capacity depreciation, implicit rental price, and investment demand for farm tractors in Canada. Can. J. Agric. Econ.,35, 373–378.

  • U.S.D.A., 1996a: Agricultural Statistics. National Agricultural Statistics Service, Washington, D.C., 511 pp. [Available from the U.S.D.A., 14th St. and Independence Ave. SW, Washington, DC 20250.].

  • U.S.D.A., 1996b: Grain: World markets and trade. Foreign Agricultural Service, Circular Series FG 3-96, Washington, D.C., 56 pp. [Available from the U.S.D.A., 14th St. and Independence Ave. SW, Washington, DC 20250.].

  • U.S. Dept. of Commerce, 1996: Statistical Abstract of the United States. 116 ed. Economics and Statistics Administration, Bureau of Census, 1022 pp.

  • Weston, J. F., and E. F. Brigham, 1990: Essentials of Managerial Finance. 9th ed. The Dryden Press, 931 pp.

  • Wilks, D. S., 1997: Forecast value: Prescriptive decision studies. Economic Value of Weather and Climate Forecast, R. Katz and A. Murphy, Eds., Cambridge University Press, 109–149.

Fig. 1.
Fig. 1.

Schematic of the relationships in the AG-GEM model.

Citation: Journal of Applied Meteorology 39, 1; 10.1175/1520-0450(2000)039<0067:DAOTIO>2.0.CO;2

Fig. 2.
Fig. 2.

Differences (with forecast − without forecast) in the Corn Belt region for debt burden ratio for the five timing scenarios.

Citation: Journal of Applied Meteorology 39, 1; 10.1175/1520-0450(2000)039<0067:DAOTIO>2.0.CO;2

Fig. 3.
Fig. 3.

Differences (with forecast − without forecast) in the Corn Belt region for present value of ending wealth for the five timing scenarios.

Citation: Journal of Applied Meteorology 39, 1; 10.1175/1520-0450(2000)039<0067:DAOTIO>2.0.CO;2

Fig. 4.
Fig. 4.

Differences (with forecast − without forecast) in the Corn Belt region for return on equity ratio for the five timing scenarios.

Citation: Journal of Applied Meteorology 39, 1; 10.1175/1520-0450(2000)039<0067:DAOTIO>2.0.CO;2

Fig. 5.
Fig. 5.

Differences (with forecast − without forecast) in national present value of producer surplus for the five timing scenarios.

Citation: Journal of Applied Meteorology 39, 1; 10.1175/1520-0450(2000)039<0067:DAOTIO>2.0.CO;2

Fig. 6.
Fig. 6.

Differences (with forecast − without forecast) in national present value of consumer surplus for the five timing scenarios.

Citation: Journal of Applied Meteorology 39, 1; 10.1175/1520-0450(2000)039<0067:DAOTIO>2.0.CO;2

Fig. 7.
Fig. 7.

Differences (with forecast − without forecast) in national present value of consumer plus producer surplus for the five timing scenarios.

Citation: Journal of Applied Meteorology 39, 1; 10.1175/1520-0450(2000)039<0067:DAOTIO>2.0.CO;2

Fig. 8.
Fig. 8.

Differences (with forecast − without forecast) in the Corn Belt region for present value of machinery and motor vehicles for the five timing scenarios.

Citation: Journal of Applied Meteorology 39, 1; 10.1175/1520-0450(2000)039<0067:DAOTIO>2.0.CO;2

Table 1.

Stages, management decisions, and approximate growth stage in the field-level decision model.

Table 1.
Table 2.

Percentage differences in yields and costs by year for producers with only historical climate information and producers using perfect climate forecast.*

Table 2.
Table 3.

Present value of changes in producer surplus, consumer surplus, and net surplus for the five scenarios in millions of U.S. dollars.

Table 3.
Save