Forecasting and Statistical Quality Assessment of Philippine Crop Production Data: Evidence from Rice and Corn
Reliable crop statistics are foundational to food-security planning, yet the literature often treats forecasting accuracy and data-quality assessment as separate tasks. This paper develops an integrated evidence synthesis around three Philippine studies that together illuminate both problems for rice and corn. The first compared Seasonal Autoregressive Integrated Moving Average and Holt-Winters models for quarterly rice and corn production and found that Holt-Winters with additive seasonality yielded lower forecast errors. The second extended the forecasting problem to machine-learning models and reported that Random Forest produced the strongest predictive performance among the tested algorithms, while performance varied across other nonlinear approaches. The third applied the Newcomb-Benford law to official crop production statistics and identified deviations in rice and corn digit patterns that warrant further validation. Drawing on official Philippine Statistics Authority documentation and broader methodological literature on forecast evaluation, survey reliability, and crop-yield prediction, the paper argues that forecastability and statistical integrity should be studied together rather than in isolation. A series can be forecastable yet still contain reporting irregularities, while a numerically plausible series can remain difficult to forecast because of structural breaks, weather shocks, or shifting production conditions. For agricultural planning, the strongest evidence base comes from combining temporal modeling with routine statistical-quality screening, transparent revision practices, and follow-up diagnostics when anomalies appear. The paper concludes by proposing a practical framework for Philippine agricultural analytics in which data integrity checks precede and accompany forecasting, thereby improving the credibility of crop outlooks used for procurement, import planning, early warning, and resource allocation.
