Update and Revision History
Farm income forecasts are updated three times each year, and historical estimates are revised as needed to reflect newly available administrative and survey data, or when estimation procedures are revised. For information on the administrative and survey data used in the Farm Income and Wealth Statistics data product, see general documentation as well as documentation on cash receipts, the balance sheet, and financial ratios. This page briefly describes the routine revision process for USDA farm income and balance sheet forecasts, and explains revisions to the historical income and balance sheet estimates when they occur.
Farm Income and Wealth—From Forecast to Estimate Across 19 Months
The periodic farm income forecasts and estimates (available in U.S. and State-level farm income and wealth statistics) published by ERS for a particular calendar year (5 releases over a span of 19 months) can vary markedly from one release to the next due to changes in the information set available to the Farm Income Team. Notably, as the year progresses, each release substitutes incrementally more 'known' information gathered via surveys and other means for forecast information that is produced by a forecast model.
Forecasts at the beginning of the year have very little known information for the calendar year under study, while forecasts occurring later are based on a growing share of known data. The first forecast is made before planting intentions are known, let alone production outcomes, which vary with weather and other factors, and price outcomes, which adjust to shifts in both domestic and global production. Farmers also respond to shifts in expected income by varying their input purchases, debt, inventories, and participation in government programs. These outcomes and decisions are revealed over the course of the production year, and are reflected in data that may lag those decisions by up to a year. Each forecast takes advantage of new information as it becomes available, and release dates are largely timed accordingly.
- The February 2016 forecast of 2016 income relied almost exclusively on information on current crop and livestock prices and was based mostly on 2015 production that would be sold in 2016, rather than production occurring in 2016 and sold that year. This is because in February very little information on 2016 production was available at the time.
- The second 2016 forecast, in August 2016, included the first survey-based information from USDA on crop production for the crop marketing year. This information on domestic crop production was supplemented by additional animal/animal product domestic production, as well as international production, consumption, and trade that allowed for an improved price forecast.
- The November 2016 release incorporated newly available price data as well as updates to current-year harvest, sales, and inventory data.
- The fourth forecast of 2016 farm income, released in February 2017, included additional 2016 monthly price information as well as marketing pattern data that enabled a more complete picture of whether crop sales occurred in one calendar year or another.
- The 2016 calendar year forecast process culminates in August 2017 when NASS estimates of State and U.S. production and expense data gathered through the 2016 ARMS are available. As of this release, the forecasts are labeled as estimates.
The staggered incorporation of additional data that become available throughout the year means that individual components of the farm income accounts are subject to varying degrees of uncertainty in any given forecast. For instance, prices change throughout the year in response to international supply and demand conditions, production estimates change as weather conditions become known, and stocks data change as producers decide on whether to store or market their production. The forecast of crop inventory adjustment is a residual of measures, taken at different points in the year, of total supply (production and beginning-of-year stocks) and use (domestic and exports). Government farm program payments—which are a function of prices, production, eligibility rules, and ad hoc disaster legislation—are also hard to predict, and forecasts improve as USDA data become available.
A historical review of forecast data published by ERS from 2001 through 2016 indicates how the four successive forecasts compare to the first published estimate, on average, over the 19-month period. The average absolute percentage difference (AAPD) indicates how the interim forecasts differ from the final estimate. Averaged over the 2001-2016 period, the calculated AAPD for the four forecasts relative to their corresponding August estimate is 11.4 percent for net cash farm income and 10.8 percent for net farm income. Also, the forecast correctly predicted the direction of the change in net cash income relative to the previous year 84 percent of the time, and net farm income 95 percent of the time. Moreover, our analysis indicates that the arrival of new data throughout a given year leads to a significant improvement in forecasts across the 19-month period, with a net cash farm income AAPD of 16 percent (February), 13 percent (August), 11 percent (November), and 8 percent (February of the following year) relative to the first estimate released in August of the following year, with corresponding net farm income AAPDs of 15 percent, 13 percent, 7 percent, and 8 percent.
The value of ERS forecasts is evident by comparing the AAPDs above to a naïve forecast AAPD. The naïve approach assumes the forecast value generated for the previous year will be the value for the current year as well. Across 2001-2016, naïve forecasts of net cash farm income had an average AAPD of 17.2 percent and the naïve forecast for net farm income had an AAPD of 21.9 percent, relative to the eventual estimated values. In sum, the naïve forecasts under-performed the USDA/ERS forecasts (with AAPD of 11.4 percent for net cash farm income and 10.8 percent for net farm income) over the 2001-2016 period.
See the data visualization, Charts of U.S. Farm Sector Forecast Evolution Over Time, which covers ERS farm income and wealth forecast accuracy across a variety of measures including net cash farm income and net farm income.
|Evolution of the ERS forecast over the forecast cycle|
|Net cash farm income ($ billion)||Forecast release||Estimate release|
|Forecast data year||February||August||November||February||August||November|
|Net farm income ($ billion)||Forecast release||Estimate release|
|Forecast data year||February||August||November||February||August||November|
|NA = Data are unavailable.
Note: Net farm income data for 2000-2012 adapted from Kuethe, Hubbs, and Sanders (2017), "Assessing the Accuracy of USDA's Farm Income Forecast." Proceedings of the NCCC-134 Conference on Applied Commodity Price Analysis, Forecasting, and Market Risk Management, St. Louis, MO.
Periodic Revisions Improve the Accuracy of Historical Estimates
The following details updates and revisions to data (available in U.S. and State-level farm income and wealth statistics) as of August 30, 2017.
August 30, 2017
With this update, the forecast for 2016 farm income becomes the first 2016 national estimate, and the first 2016 State estimates become available. The update also revises the initial 2017 forecast done in February 2017. The 2016 National and State estimates reflect newly available data on cash receipts from NASS and preliminary data from the 2016 Agricultural Resource Management Survey (ARMS).
Relative to the previous estimates released in February 2017, some 2014 and 2015 estimates—including production, sales, and percent of sales marketed by month from NASS, as well as macroeconomic variables sourced elsewhere—were revised to incorporate newly available data. The new information from NASS increased cash receipts and affected inventory adjustments in the income statement, as well as inventories carried in the sector balance sheet. Combined, total value of production increased by $59 million (0.01 percent) and $669 million (1.1 percent) for 2014 and 2015, respectively, as a result of these newly available data. Inventory values in the balance sheet decreased by $5.0 million, on average, in 2014 and 2015, or 0.003 percent of previously reported crop inventory data.
The macroeconomic variable changes affected estimated net rent received by farm operators, lowering it by $2 million in 2014 and increasing it $0.09 million in 2015. Interest expenses for 2015 only were revised down $28 million, or 0.17 percent relative to previous estimates, and capital consumption was revised upward by $233 million and $200 million, or 0.47 and 0.48 percent of their previous values for 2014 and 2015, respectively.
Upward revisions to net farm income due to newly available data in 2014 and 2015 were $163 million (0.18 percent) and $497 million (0.62 percent), respectively.
In addition to changes to 2014 and 2015 estimates, mink cash receipt data for 2008-2011 were revised upward by $61 million per year, on average (36-percent increase relative to previous data) due to corrections in underlying data. This change increased total sector-level cash receipts by 0.02 percent, on average, across the 4 years. Finally, capital consumption in 2013 was revised downward by $28 million, or 0.07 percent of the previous capital consumption value.
These changes in previously reported data also resulted in slight upward revisions to net farm income in each of the years that they occurred. Net farm income, due to correction of previously reported data, is revised upward by $68 million (0.09 percent) for 2008, $69 million (0.11 percent) for 2009, $46 million (0.06 percent) for 2010, $59 million (0.05 percent) for 2011, and in $27 million (0.02 percent) for 2013.
With this release, nominal values are adjusted for inflation using the chain-type GDP deflator, base year=2017. Database files, in CSV file format, were added for previous data releases made in August 2014, November 2014, February 2014, and August 2015.
March 8, 2017
On March 8, 2017, the Excel file "U.S. farm sector financial indicators, 2011-2017F" was reposted to include estimates of farm debt by lender that were inadvertently labeled as NA in 2014 and 2015.
February 8, 2017
The 2015 cash receipts estimate is revised based on updates to data released by NASS on crop production and monthly crop marketing data. These changes, which also had minor effects on expenses and asset values, were slight, as net farm income for 2015 is revised downward to $80.875 billion—a $3 million decline (less than a hundredth of a percent) relative to the November estimate.
Also, the farm business average net cash income report for 2010 to the present has been further disaggregated to show additional detail. Farm businesses specializing in cotton and soybeans are now displayed separately and the 'other crops' farm specialization entries are redefined to include peanuts, rice, mixed grains, and other field crops. The net cash farm income estimates for farm businesses now all reflect the most recent ARMS data, going back to 2010.
Finally, the inflation-adjusted values are revised for all data from 2013 through 2016, reflecting revisions to the chain-type price index (GDPCTPI) maintained by the Bureau of Economic Analysis, available on the St. Louis Federal Reserve website.
Comparisons to November forecast for fiscal year 2016. The calendar year forecast for net farm income in 2016 has been updated to reflect new information, resulting in a small (2.0%) increase in the income forecast relative to November. Primary drivers of the increase were crop and livestock receipts. The graphic below shows the difference in individual components of the farm income forecast for 2016 relative to the one released on November 30, 2016.
Compared to the November 2016 forecast for 2016 equity and assets, the February outlook has slightly improved. As a result, 2016 farm sector solvency ratios forecast in February 2017 are projected to strengthen slightly relative to their November 2016 predictions.
November 30, 2016
The November 2016 release of the 2016 farm income forecast is the second revision since the initial February forecast. The revision incorporates updated data from NASS, including newly available monthly marketing percentage data. Due to a revision by NASS, 2012 cucumber data for Missouri have been removed and miscellaneous crops data between 2008 and 2012 have been revised due to a prior omission. Additionally, ERS changed the way monthly prices are calculated for States/commodities for which price data are not available, with a minimal effect on commodity marketing-year cash receipt estimates for those crops. The change did not affect the 2016 farm income forecast but it affected the estimates for 2015 and previous years. The table below compares the 2015 estimates published in August 2016 for net farm income with the estimates published in November 2016 and reflects the cumulative impact of all of the changes described above. The changes to net farm income range from a .06-percent decline to a 0.17-percent increase. Changes to farm sector equity range from a 0.02-percent decline to a 0.05-percent increase.
Beginning with the November 2016 forecast, an additional sector-level balance sheet is being published, which presents assets and debt by current and noncurrent component. The ERS is also now publishing three additional financial metrics—current ratio, working capital ratio, and working capital-to-gross-revenues ratio—recommended by the Farm Financial Standards Council.
A new line in the Federal Government direct farm program payments report has been added to report payments for a new program called the Cotton Ginning Cost-Share (CGCS) program. The program provides cost-share assistance payments to cotton producers with an ownership share in the 2015 cotton crop. The program was established under the statutory authority of the Commodity Credit Corporation Charter Act. The sign-up period for the program was June 30, 2016, through August 5, 2016. Eligible producers receive in 2016 a one-time cost-share payment based on their share of 2015 cotton acres reported to USDA’s Farm Service Agency multiplied by their regional payment rate. There are 4 regional payment rates equal to 40 percent of the average ginning cost for that production region. Cost-share payments are capped at $40,000 per individual or entity.
August 30, 2016
With this update, the forecast for 2015 farm income becomes the first 2015 national estimate, represents the release of the first 2015 State estimates, and revises the initial 2016 forecast done in February 2016. The 2015 national and State estimates reflect newly available data on cash receipts from NASS and preliminary data from the 2015 Agricultural Resource Management Survey (ARMS). New information from the 2014 Tenure, Ownership, and Transition of Agricultural Land (TOTAL) survey is used to estimate net rental income to operator and nonoperator landlords for years 2014 forward. Cash receipt data for 2013 and 2014 were revised to incorporate newly released and revised data—including production, sales, and percent of sales marketed by month—from the National Agricultural Statistics Service.
Additionally, production expenses in 2011 were revised to incorporate updated Federal insurance premium data from USDA’s Risk Management Agency. Farm asset value and debt data in 2013 and 2014 were updated in order to incorporate changes in underlying ARMS data since the last release, as well as changes to debt data collected from administrative sources.
Beginning with the August 2016 forecast, Federal commodity insurance indemnity and premium data are now reported as a subcomponent of total commodity insurance indemnities and premiums. Additionally, a new method is used for incorporating World Agricultural Supply and Demand Estimates (WASDE) marketing-year price forecasts for corn, soybeans, wheat, sorghum, barley, oats, rice, and cotton into the ERS calendar-year cash receipts forecast.
May 2, 2016
On May 2, 2016, the Agricultural Resource Management Survey data—the data source underlying the two tables Farm-level average net cash income by farm typology and sales class and Farm business average net cash income by commodity specialization and region—were revised to incorporate data quality improvements for 2012-2016F and to correct a programming error that inadvertently excluded cash sales of hay and sugarcane/sugarbeets from ARMS crop cash income in 2013 and 2014. Correction of the programming error increased the reported average farm-level net cash income by $3,100–$3,500 per farm in each revised year 2013-16F (about 7 percent per year). For farm businesses, the increase in average per-farm net cash income was $6,300-$6,700 (5-6 percent). The farm sector-level forecasts and reports were not affected.
February 9, 2016
Final revision of 2015 forecast, using updated data form NASS and other administrative sources. Estimates from 2008-14 were revised to account for updated NASS data, and revised 2012 Agricultural Resource Management Survey (ARMS) data. The first forecast for the 2016 farm income and balance sheet is issued. Net farm income for 2012 is revised upward by $2.9 billion due to a change in the calculation of the gross imputed rental value of dwellings. The revision affects gross imputed rental value of dwellings and net farm income reported for the U.S. and each State. The revision does not affect the income reported for other years. Also, the 2012 estimates in the two tables Farm-level average net cash income by farm typology and sales class and Farm business average net cash income by commodity specialization and region were revised downward to correct a miscalculation of gross income from production of 'other livestock.' The correction only affects average cash income estimates on these two tables; sector level estimates of cash receipts, net cash farm income and net farm income were not affected. The correction resulted in a downward revision of $4,600 (4.1 percent), to the 2012 estimate of farm business average net cash income. The 2012 average net cash income estimates by region, commodity specialization, sales class and typology were also revised downward, by varying amounts.
November 24, 2015
Revision of 2015 forecast. Data from 2008 forward were revised with more recent data from NASS to account for revisions to sheep and lamb inventory data, more recent citrus and potato data, new commodity monthly marketing percentages data, and revisions to the 2013 and 2014 Agricultural Resource Management Survey (ARMS) data.
In addition, on November 24, 2015, ERS issued an errata to reflect corrections to the source State cash receipts data for the "all other animals and products" and "miscellaneous crops" categories for 2008-12 made by the National Agricultural Statistics Service for California, Delaware, Connecticut, Massachusetts, Louisiana, and Nevada. Changes to these States’ cash receipts estimates are reflected in U.S. total cash receipts estimates for 2008-12, as well as all of the income measures that use cash receipt data as a component (including net farm income, net cash income, value added, and farm-level and farm business average net cash income). The changes added 0.6 to 0.9 percent to U.S. total cash receipts in each of the years from 2008 to 2012.
October 14, 2015
Errata: On October 14, 2015, the report for Annual cash receipts by commodity, U.S. and States, 2008-2015F was reposted. An original version of this report, posted August 25, 2015, did not return inflation-adjusted dollar values when "real" was selected as an option; nominal values continued to be displayed. This option now provides correct values. This was the only report in the set that was affected.
August 25, 2015
Release of 2014 final estimates, 2014 State estimates, and revision of the 2015 forecast. Data from 2008 forward were revised to account for changes in Agriculture Census weights, Census values and a change in estimation and forecast methodology.
February 10, 2015
Release of the 2015 forecast and revision of the 2014 forecast. For 2012-13, insurance premiums and indemnities (including crop insurance premiums and indemnities) are reported as separate income and expense items. Prior to this release, and for all other years, insurance indemnities were included in other farm income and premiums were included in miscellaneous expenses.
December 12, 2014
Errata: On December 12, 2014, the ERS farm income estimates for 2008-2013 and the 2014 forecast, released on November 25, were revised to correct coding and data input sourcing problems in the underlying farm income database. Revisions also incorporate new data that became available following the November 25 release. While changes to individual State level estimates may be larger, at the U.S. level, these corrections and data revisions increased 2013 net farm income by 2 percent, to $129 billion. Also affected were the value of year-end inventories and related measures in the balance sheet, as well as production expenses, total value of production, gross and net value added, the value of inventory change, and their respective crop components in the income statement. Net farm income forecast for 2014 increased 0.4 percent, to $97.3 billion.
November 25, 2014
Beginning with the release of the 2012 Census of Agriculture in May of 2014, NASS announced the preparation and release of historical revisions to a number of important statistical products that would occur throughout the year. Most of these estimates are used in preparation of the farm sector value added and net income measures produced by ERS (see Documentation to the data product). The reports release included:
- Crop Values
- Field Crops
- Potatoes and Sweet Potatoes
- Rice Stocks
- Stocks of Grains, Oilseeds, and Hay
- Noncitrus Fruits and Nuts
- Meat Animals Production, Disposition, and Income
- Milk Disposition and Income
- Poultry Production and Value
- Chickens and Eggs
- Hogs and Pigs
- Milk Cows and Production
- Sheep and Goats
- Citrus Fruits
- Agricultural Land Values and Cash Rents
- Farms and Land in Farms
- 2012 Census of Agriculture
The result of incorporating the historical revisions for crop and livestock cash receipts was a relatively small reduction in livestock cash receipts (-0.18 to -1.33 percent) and a similar small percentage increase in crop receipts over the 2008-12 period (0.58 to 5.14 percent). Price and output revisions also resulted in changes for the value of crop and livestock production and estimated inventory changes. As a result, total value of agricultural production was revised downward by between 3-5 percent during 2008-12, with the largest reduction occurring in 2011. Changes to revenues from services and forestry (includes forest products sold, gross imputed rental value of farm dwellings, machine hire and custom work, and other farm related income) were also relatively minor for 2008-11. Forest products sold were the main component that changed, in the range of -3.41 to 1.12 percent. During 2012, all items in this category were updated to census values. There were no revisions to expenses during 2008-11.
August 26, 2014
Farm sector real estate asset values from 2008 to 2012 were revised in order to incorporate final estimate information from NASS on land values and land in farms. The value of machinery and motor vehicles assets was revised in 2012 to incorporate newly available census of agriculture information. Reflecting these changes, historical farm balance sheet values for farm sector equity (assets minus debt), and the sector debt-to-asset, debt-to equity, and equity-asset ratios were revised for data covering 2008-12.
March 11, 2014
On March 17, 2014, farm debt for 2002-06 in the Farm Sector Balance Sheet Table (located in Farm Income and Wealth Statistics) was revised to correct a programming error. As part of the February 2014 release, debt held by Farmer Mac at the end of each year for 2002-06 was added to the table for the first time, but was not reflected in calculations of total real estate debt or total farm sector debt. As a result, farm sector equity (assets – debt), and the sector debt-to-asset, debt-to equity, and equity-asset ratios were also incorrect from 2002-06. All series were revised to provide correct data.
February 11, 2014
Revisions were made to two data series since the November 2013 release. The amount of real estate debt held by Farmer Mac was previously shown as "not available" for 2002-06. However, as new information and analyst time became available, it was possible to provide these data. Additionally, a rounding error was corrected in the Net Farm Income series for 1950-59.
November 26, 2013
Following the August 2013 farm income and wealth statistics data release, errors were detected in several series. Many of the errors impacted national data, as well as data for each State. This is true for intermediate production expenses and labor expenses (2012), interest expenses (2008-12), real estate interest expenses (1988), nonreal estate interest expense (1971-80), and net rent to landowners (1971-80). Additionally, the value of inventory change for meat animals (2010-11) and real estate taxes excluding operator dwellings (1910-48) were found to include errors affecting only national (not State) data. The November 2013 release included corrected data.
August 27, 2013
Revised farm balance sheet estimates for 2002-11
Following the February 2013 update, a thorough review of the estimation process for USDA’s farm sector balance sheet was undertaken, and several changes were instituted to more accurately reflect farm sector debt. Estimation procedures were amended to improve handling of nonresponse to debt-related questions in the Agricultural Resource Management Survey (ARMS) and to recognize changes in the lending industry and associated administrative data over time.
Administrative data from all the major agricultural lenders are routinely adjusted using the most recent ARMS to estimate the portion of agricultural debt used for farm purposes. Comprehensive procedures were instituted to impute missing data in ARMS (due to respondent nonresponse) from 2002 to 2011, and these revised data were applied to estimate the share of lender-reported agricultural debt that was used for farm business purposes. Debt attributed to individuals and other lenders who do not routinely report farm lending activities was also re-estimated based on revised ARMS data. Finally, estimation procedures were amended to reflect Farmer Mac’s changing role in the farm debt market and to account for other changes in the availability of administrative data on agricultural lending. The impact of these changes was sizeable in some cases.
Revised farm income estimates for 2008-11
The above-mentioned revisions to the farm balance sheet estimates affected estimated interest expenses for farm operations. These changes, in turn, affected estimates of net farm income. Revisions are ongoing, but the August 2013 release incorporates revisions to interest expenses and income estimates for 2008-11.