The standard economic analysis of product-line pricing by Mussa and Rosen (1978) implies that higher-quality varieties command higher absolute mark-ups. It is widely claimed that this property does not apply to wine lists. Restaurateurs are believed to overprice the second-cheapest wine to exploit naïve diners embarrassed to choose the cheapest option. This paper investigates which view is
correct. We find that the mark-up on the second cheapest wine is significantly below that on the four next more expensive wines. It is an urban myth that the second-cheapest wine is an especially bad buy. Percentage mark-ups are highest on mid-range wines. This is consistent with the profit-maximising pricing of a vertically differentiated product line with no behavioral elements, although other
factors may contribute to the price pattern.

That is from a new paper by David de Meza and Vikram Pathania, via The Browser (which is worth paying for).