|
Abstract:
|
Small area models have gained increased attention by statistical agencies. They can “borrow strength” from related areas across space and/or time or through auxiliary information and they can provide “indirect” but reliable estimates for small areas with small or even zero sample sizes while also increasing the precision. The United States Department of Agriculture’s (USDA’s) National Agricultural Statistics Service (NASS) conducts the Cash Rents Survey (CRS) to provide the basis for county estimates of the cash rent paid for irrigated cropland, non-irrigated cropland, and pasture land. Estimates of cash rental rates are useful to farmers, economists, and policy makers. However, realized sample sizes at the county level are often too small to support reliable direct estimates. To improve the less reliable direct estimates, model-based estimates have been extensively discussed in the literature. We propose a hierarchical Bayesian (HB) area-level two-component mixture model to account for outliers that incorporates two years of data with a discounting factor for the first year. When compared to the standard HB method based on normality assumptions, the proposed method to handle outliers is robust. In addition, it is a general model that puts the two years of data together and it avoids correlations by using a power prior that partly discounts past data. A 2016 and 2017 case study illustrates the improvement of the direct survey estimates for areas with small sample sizes by using auxiliary information and by borrowing information across areas.
|
