In their shopping mall data, anchor stores are 60% of the floor space. But they pay $4.13 rent per square foot, against $29.37 for the non-anchor stores. Meanwhile average sales per sq foot are $185.34 adn $317.68. All this is in table 1 of their paper, below.
Is this sensible? Let's assume the Mall owners want to maximise total sales in the Mall (they will want to maximise their profits, but suppose they are related to total sales), and they have some spare space. What would a price discriminating monopolist do? Answer: equalise marginal revenue in each sub market (for constant marginal costs). In this context, this means setting rents so that an extra square foot rented to an Anchor generates the same sales to that rented to a non-Anchor.
It looks like they are failing, see Table 1 since the non-Anchors generate much more sales. So what is happening? It's like that the presence of the anchors adds demand to the non-Anchors. So another anchor square foot does not just generate $185 but an additional effect on the non-Anchors. Their statistical analysis, using variation in sales and presence of different store types, suggests another Anchor square foot generates additional sales for non-Anchors of $116 per sq ft. Thats a total of 185+116=$301, which is very close to $317. The externality is internalised.