Mixed

FAO food balance sheet estimates of national per capita food supply substantially overestimate actual individual-level dietary intakes for most food groups (e.g., by 74.5% for vegetables and 270% for whole grains) and underestimate others (e.g., by 250% for beans/legumes), rendering raw FAO data invalid for assessing individual dietary quality or diet-disease relations without calibration.

Do not use UN FAO food balance sheets to judge how much people actually eat. These numbers are systematically wrong, often overstating consumption by 50-270% for foods like vegetables and grains. If you must use FAO data for global comparisons, you must apply specific calibration equations that account for country, age, sex, and region to correct for this bias.