Leah Brooks: When a store owner finds his street is dirty, he is faced with a vexing problem. He can clean just the street in front of his store for a minimal cost; this is unlikely to please his customers who need to walk in front of other stores to reach his store. He could clean the entire street, but that is likely prohibitively expensive. He would be willing to clean in front of his store if all the other owners would clean in front of theirs, but he does not trust the others to clean reliably. The other owners similarly desire cleaner streets but also lack the means of enforcing a binding agreement. This collective action problem is precisely what a business improvement district (BID), with its powers of coercive taxation, is designed to solve. By providing a mechanism allowing owners to commit credibly to the provision of public service, owners become willing to invest in local public goods.
BIDs' presence has grown enormously over the last forty years. The city of New York had no BIDs in 1976 and has fifty-five today; Los Angeles had no BIDs in 1994 and has more than thirty today.1 Indeed, roughly half of California cities with a population over 25,000 have at least one BID.2 With this great growth in the number of BIDs, it is natural to ask whether they succeed in their goal of resolving collective action problems in commercial neighborhoods. If BIDs are successful, we would expect their benefits to be capitalized into property prices.
Studying BIDs in New York City from 1984 to 2002, Ellen, Schwartz, and Voicu find that BIDs do increase property values: prices increase roughly 15 percent after adoption. This increase is driven mostly by larger, predominantly office space, higher-spending BIDs; residential properties inside and near to BIDs are largely unaffected, and BIDs cause a one-shot increase, not a change in trend. [End Page 32]
I focus on three aspects of this project: the authors' ability to measure cleanly the impact of BIDs, where BIDs fit into our understanding of municipal policy, and, taking the even longer view, where BIDs fit into our broader understanding of the provision of local public goods.
Ideally, we would compare the performance of neighborhoods with BIDs to those same neighborhoods in a parallel universe without any BIDs. As this is not feasible as a practical matter, and the matter of BID impacts is of important consequence, the authors turn to regression adjustment techniques as a second best. In the authors' first specification, we may interpret the estimate as the causal impact of BIDs if we believe that BID and non-BID areas differ only in zip code–level fixed characteristics and property-related characteristics as measured in 1999 and that BID adoption is not taken at a particular time-point in the path of commercial property prices. This is a very restrictive set of assumptions, and the paper would be improved with more evidence—summary statistics—as to whether they are believable.
For example, suppose that BIDs consist primarily of dense commercial areas, and non-BIDs consist of less dense commercial areas. If the trends in pre-BID prices differ for these two types of areas, the specification will attribute some of this difference in pre-BID trend to the effect of the BID. Or, if BIDs form due to the presence of a local organizer, and the local organizer could have effects that predate the BID, the estimator would similarly attribute to the BID some of the effects that should be attributed to the organizer.
If BIDs were identical to non-BIDs save for the control characteristics, we would not expect the coefficient on ever being in a BID (as opposed to being in a BID after BID adoption) to significantly explain property price. However, the coefficient on this variable is persistently significant, which suggests that BIDs do differ in significant ways, even before adoption...