We study how substitutability between clean and dirty alternatives affects the effectiveness of environmental regulation in a field experiment that controls for the choice set of respondents. We consider four product categories with clean and dirty alternatives: (i) cola products in plastic bottles vs. in aluminum cans; (ii) skimmed vs. whole milk; (iii) chicken meat vs. beef meat; and (iv) margarine vs. butter. We employ two neutrally framed treatments to quantify the willingness to substitute between clean and dirty alternatives in each product market, namely a change in relative prices and the removal of the dirty alternative, leaving respondents the option of buying one of the remaining clean alternatives or nothing. We then compare the impact of a carbon footprint label and a Pigovian subsidy to the clean alternatives. While both instruments increase the market share of the clean products, their impact is higher when clean and dirty alternatives are close substitutes. We also find evidence that motivation crowding is present and increases with substitutability. Our results highlight the importance of product markets in the design of consumer-orientated policies.