Notes: The table presents parametric regression discontinuity estimates of the treatment effect of broadband use on 1-year change in log employment. Data period 2000–2004. We choose not to report the coefficient estimates for the polynomial functions and the other control variables (firm age, foreign ownership and a multi-plant dummy) for the sake of parsimony. Industries are measured at the three-digit SIC level and regions refer to government office regions. The polynomial function is chosen according to the specification that minimizes the Akaike Information Criterion. The first-stage results refer to the effect of being attached to an ADSL broadband-enabled telephone exchange on the likelihood of a plant using broadband. Broadband boundary is a zero-one indicator equal to one if the plant is inside the area of broadband-enabled telephone exchanges. In Columns 3 and 4, two treatment effects are estimated, interacting Broadband adoption with a rural dummy and an urban dummy, and the Boundary instrument is similarly interacted. ***, ** and * indicate significance at the 1, 5 and 10% level, respectively, with standard errors clustered at the firm-level reported in parentheses.