Prevented planting payments

The prevented planting provision became a standard component of US crop insurance in 1994 with the purpose of decreasing the need for ad hoc disaster assistance for producers. Producers with policies eligible for prevented planting indemnities, which include Revenue Protection (RP), RP with the Harvest Price Exclusion (RPHPE), and Yield Protection (YP) insurance plans, have several options if they are unable to plant within their designated window. However, the producer has 72 hours after the final planting date to provide a notice of loss to their insurance agent and decide which prevented planting option to pursue. These options include planting the originally insured crop during a late planting period, but the producer’s guaranteed coverage reduces 1% each day during the late planting period. The producer could also choose to plant a different insured crop after the late planting period. For instance, planting corn could be switched to growing soybeans since the soybean planting window is later than corn. This could also be an uninsured alternative crop like an annual grass for haying and/or grazing, which also provides a partially prevented planting indemnity payment.

However, the most selected option is to take full prevented planting indemnity (USDA OIG 2013). The USDA OIG (2013) report found more than 99% of all prevented planting claims selected this option. This option pays a percentage of the insurance coverage guaranteed amount and restricts the producer from planting a harvestable crop. The prevented planted field would either need to be left fallow or could be planted to an unharvested cover crop. A producer selecting this option over planting late in the year or planting an alternative crop is what Kim and Kim (Reference Kim and Kim2018) defined as ex post moral hazard.

Mathematically, the net returns from the full prevented plant payment option is defined as

(1) $$NR_{ik}^{FP} = p_i^Gy_i^G{\delta _k}{\theta _i} - C_i^{bp} - {I_{ik}}$$

where $NR_{ik}^{FP}$ is the net returns to the full prevented planting indemnity payment ($/acre) ith crop (i = corn or soybeans) with kth crop insurance coverage level; $p_i^G$ is the guaranteed price for the insurance policy; $\;y_i^G$ is the guaranteed yield or actual production history (APH); $\;{\delta _k}$ is the insurance coverage level and is equal to the value of k (k = 50%, 55%, 60%, 65%, 70%, 75%, 80%, or 85%); ${\theta _i}$ is the prevented planting coverage factor, which for corn is 55%, and soybeans is 60%; $C_i^{bp}$ is the expected pre-planting production cost ($/acre); and ${I_{ik}}$ is the producer portion of the crop insurance premium ($/acre). For example, a corn producer with an RP policy, with 75% coverage level, an APH of 150 bu/acre, and projected price of $3.86/bu would have a guaranteed revenue minimum of $434.25/acre ($434.25 = $3.86 × 150 × 0.75). Prior to planting, assume the producer has spent $100/acre on the insurance premium, land, machinery, and chemical. The full prevented planting payment would pay 55% of the guaranteed coverage amount, which is $238.84/acre ($238.84 = $434.25 × 0.55) resulting in a profit of $138.84/acre ($238.84 − $100). Building from this example, if the coverage level was 85%, the guaranteed revenue minimum is $492.15/acre ($3.86 × 150 × 0.85) with a prevented planting payment of $270.68/acre and net profit of $170.68/acre. Producers’ full prevented planting payment increases as the guaranteed coverage level of the policy increases (Boyer and Smith Reference Boyer and Smith2019; Kim and Kim Reference Kim and Kim2018).

Net returns from the full prevented planting payment are a function of the insured unit’s state-specific guaranteed price and farm-specific insurance premiums, APH yield, and pre-plant costs. Thus, it is plausible that prevented planting indemnities will vary geographically. A substantial portion of these pre-plant costs will be land rents. Land rents vary widely across US soybean and corn acres and are a function of cost structures, yield potential, government payments, and land use and amenities (Allen and Borchers Reference Allen and Borchers2016; Kirwan Reference Kirwan2009). Regions with higher yields will have potential for higher prevented planting indemnities but will likely have higher land rents (Kirwan Reference Kirwan2009; Paulson and Schnitkey Reference Paulson and Schnitkey2013). This variation in cost structure across regions could also influence payment disparities.

To solve for a prevented planting coverage factors that would provide an equal payment to pre-plant costs across insurance coverage levels and regions, we set equation equal to zero and solve equation (1) for the prevented planting coverage factor. The net returns to full prevented planting payment could also be set to some minimum payment $\left( M \right)$ , which might be projected returns to a secondary crop. Mathematically, this can be expressed as

(2) $$\theta _{ikcm}^* = {{C_{ic}^{bp} + M} \over {p_{icm}^Gy_{ic}^G{\delta _k}}}$$

where $\theta _{ikcm}^*$ coverage factor for the coverage plan m (RP, RPHPE, YP) with kth coverage level that is specific to county c, and $M$ is a selected minimum payment provided to the producers. This calculation would capture regional and policy variation in the prevented planting coverage factor that would accomplish the original purpose of this provision to protect producers from failed planting by compensating them for their pre-plant costs or some established revenue minimum. The premium cost was removed from the prevented planting coverage factor since insurance does not reimburse the premium but the losses.

Data

Data were collected from USDA RMA Summary of Business database from 2011 to 2020 for corn and soybeans in Arkansas, Kentucky, Illinois, Indiana, Iowa, Louisiana, Mississippi, Missouri, Ohio, and Tennessee (USDA RMA 2021c). The corn and soybean acres in regions near major river basins commonly make the majority of prevented planting acres and are the most frequently designated as prevented planting acres due to excessive moisture (USDA OIG 2013; USDA Farm Service Agency [FSA] 2021; USDA RMA 2021b; Wu, Goodwin, and Coble Reference Wu, Goodwin and Coble2020; Boyer, Park, and Yun Reference Boyer, Park and Yun2023). Therefore, we focused on the Mississippi River Basin states given this region is primarily corn and soybean acres to explore if regional variation exists.

These county-level data include the number of insurance policies sold, policies indemnified, acres coverage, total premiums, subsidies, and indemnity payment by county, state, year, coverage plan, and coverage level. For example, in a specific county there could be five observations in a year for RP policies with 50%, 55%, 65%, 75%, and 80% coverage level. There could also be a similar five observations within that same county for other coverage plans like YP. Thus, each county could have multiple observations within a year. The projected price data were gathered from the USDA RMA Price Discovery database for 2011 to 2020 (USDA RMA 2021d). These are state-level prices set by USDA RMA. APH yield data are not publicly available, which is a challenge for researchers who analyze crop insurance policies; thus, studies typically use USDA National Agricultural Statistical Service (NASS) yields (Kim and Kim Reference Kim and Kim2018; Seo et al. Reference Seo, Kim, Kim and McCarl2017).

County-level cost of production data is difficult to obtain and hard to estimate. We first collect county-level cropland rent from USDA NASS by year. Pre-plant costs typically include land rent along with chemical and machinery costs for burndown and pre-emerge herbicides (Boyer and Smith Reference Boyer and Smith2019).Footnote ² Data on county-level chemical and machinery costs do not exist; thus, we use POLYSYS budgeting system to generate county-level pre-plant costs for chemical and machinery. POLYSYS is a partial equilibrium socioeconomic simulation modeling system of the US agricultural sector in which production decisions are made. POLYSYS is a system of interdependent modules simulating crop production, national crop demand and prices, and livestock supplies and demand. POLYSYS also generates cost of production information for various crops using the 13 USDA Economic Research Service regional budgets for each crop and tillage combination. Budgets are estimated for all counties in our study using “inverse distance weighting” interpolation for costs and input quantities (Hellwinckel Reference Hellwinckel2019). We use POLYSYS budgeting system to generate pre-planting production costs for corn and soybeans at the county level within the region of study. POLYSYS generates pre-plant costs for 2020, and these costs are adjusted using producer price indices used in USDA baseline projections to estimate pre-plant costs in prior years. These POLYSYS pre-plant data had minor variation at the county level, but the land rent costs accounted for most of the total pre-plant costs and pre-plant cost county-level variation. Table 1 shows the summary statistics of yields and pre-plant costs.

Table 1. Summary statistics of the estimated yields and POLYSYS generated pre-plant costs from 2011 to 2020 for corn (n = 63,989) and soybeans (n = 72,353)

¹ United States Department of Agriculture National Agricultural Statistic Service.

Data collected were substituted into equation (2) to estimate the prevented planting coverage factor by crop, tillage system, county, coverage level, and coverage plan. We chose to set the minimum payment to be zero for this study, which means the indemnity payment would cover the pre-plant costs and not exceed this amount. Changing this value would result in same relative changes between regions and policies but would change the absolute value. Table 2 shows the summary statistics of the prevented planting coverage factor for corn and soybeans by tillage at various coverage levels. The table demonstrates how the coverage factor varies across coverage levels with coverage factors being higher at lower coverage levels and declining as the coverage increases. This is because the prevented planting payment increases as the coverage level increases.

Table 2. Summary statistics of the estimated coverage factor from 2011 to 2020 for corn and soybeans planted with no-tillage and conventional tillage

Note: Coverage factors were calculated using Equation (2).

Estimation

The estimated prevented planting coverage factor becomes the dependent variable of our estimation to measure the marginal impacts of insurance products’ features, such as coverage level and type of plan, on the estimated prevented planting coverage factor. We estimate four pooled ordinary least squares regression models with estimated prevented planting coverage factors for corn no-tillage planting, corn tillage planting, soybean no-tillage planting, and soybean no-tillage planting as dependent variables. The independent variables included coverage level, coverage plan, state, county, year, and state-specific time trend dummies. We estimate the models specified as each of the four dependent variables:

(3) $$\eqalign{ & \theta _{ckm}^* = \alpha + \sum\nolimits_{k = 1}^{K - 1} {{\beta _k}} C{L_k} + \sum\nolimits_{m = 1}^{M - 1} {{\gamma _m}} {G_m} + \sum\nolimits_{c = 1}^{C - 1} {{\varphi _c}} {D_c} + \sum\nolimits_{s = 1}^{S - 1} {{\omega _s}} S{T_s} \cr & \quad \quad \quad + \sum\nolimits_{t = 1}^{T - 1} {{\mu _t}} T{D_t} + {\xi _s}\left( {{t_s}} \right) + {\varepsilon _{ckm}} \cr} $$

where $\theta _{ckm}^*$ is the prevented planting coverage factor; $C{L_k}$ is a dummy variable for the kth insurance policy coverage level; ${G_m}$ is policy type dummy variable; ${D_c}$ is a county dummy variable for county $c$ ; $S{T_s}$ is a state dummy variable for state $s$ (s = 1,…,S); $T{D_t}$ is a year dummy variable at year $t$ (t = 1,…,T); ${\xi _s}\left( {{t_s}} \right)$ is the state-specific time trend; ${\varepsilon _{ckm}}$ is the error term; and $\alpha, \beta, \gamma, \;\varphi, \;\omega, \mu \;$ and $\xi $ are parameters to be estimated. The models were estimated with Ordinary Least Squares with Huber-White Sandwich estimator.

The state dummy variable is expected to capture state-level prices. The state-level time trend variables were included to absorb any long-run technological changes that might impact planting, such as equipment, installation of drainage tiles, or biotechnical advancements across states. The county dummy is expected to capture some other unobserved county-level factors. The year dummy controls for year-to-year variable in variables such corn and soybean prices. While we include these in the model, changing prevented planting coverage by year could cause confusion across producers and insurance agents. The results for year-to-year variability in the prevented planting coverage factor are not discussed.