Although most environmental regulations in the U.S. are founded on federal statutes administered by the U.S. Environmental Protection Agency, state environmental agencies have sizable discretion. States allocate air pollution cleanup across plants, administer permits for water pollution, and do most air and water pollution inspections. The result is (possibly) sizable differences in stringency across states. Some variation is deliberate, with plants in dirty areas being under greater regulatory pressure to reduce emissions. Other variation may be due to political differences across states in their preferences for environmental protection versus economic development.

This paper focusses on the allocation of production across plants owned by the same firm, using data from the steel, oil, and paper industries. If stricter regulation leads to higher production costs, firms should shift production away from stricter states. In prior work, we found that fewer new plants are established in states with stricter regulations. We also found that plants with higher levels of pollution abatement costs tend to have lower levels of productivity. Thus differences in environmental regulation may be large enough to influence firm decisions.

We use plant-level information from the Census Bureau's Longitudinal Research Database (LRD). The LRD includes information at five-year intervals from the Census of Manufactures for all U.S. manufacturing plants. The LRD links each plant's data over time, and identifies plants owned by the same company. We use this plant-level data for large multi-plant companies in each industry to measure how much of each company's production takes place within each state. Using LRD data from 1963 to 1992, we can identify shifts in production before federal environmental regulations took effect, as well as shifts during the regulatory period.

We examine six measures of state-level differences in regulatory stringency, taken from our earlier research. These measures include political support for environmental regulation (congressional voting patterns; state membership in environmental organizations), spending on environmental regulation (state regulatory spending; manufacturing abatement costs), and direct stringency measures (state-specific environmental laws; numbers of enforcement actions and inspections in the state). We also include other state-level variables that might be expected to affect production decisions, including changes in population and the location of other industries that purchase the products of these industries. Our analysis tests for an impact of state-level regulatory differences on the location of production.

We use EPA regulatory databases to identify the compliance status at plants owned by each company. We differentiate between 'complying' and 'non- complying' firms (based on their plants' compliance status), to test whether the two types of firms respond differently to state-level regulatory differences. The direction of the impact is uncertain: firms that hate the rules may both break them and flee them, but firms that always plan to comply face the biggest differences in abatement costs across states, while disobedient firms may not plan to abate their pollution, even in strict states (possibly avoiding some costs). These results may help guide regulatory agencies in allocating enforcement, if they help identify firm characteristics that are associated with persistent violations of environmental laws.