This paper pro­vi­des an ori­gi­nal third-step iden­ti­fi­ca­ti­on stra­tegy using ins­tru­men­tal vari­a­bles to eva­lu­a­te the cau­sal impact of highway invest­ments on the local eco­nomy. First, we cons­truct a novel nati­o­nal highway data­set at the muni­ci­pal level in Bra­zil using the Growth Acce­le­ra­ti­on Pro­gram (PAC) (2007–2018) as a case study. Second, we rely on some of the main infras­truc­tu­re pro­ject costs to pro­po­se seve­ral cos­tre­la­ted ins­tru­ments to cor­rect for mea­su­re­ment errors in the road vari­a­bles. Third, we cir­cum­vent the omit­ted vari­a­ble bias from the non-ran­dom pla­ce­ment of roads by buil­ding ins­tru­ments based on glo­bal cost mini­mi­za­ti­on methods, his­to­ri­cal plans, and the pro­pen­sity of a muni­ci­pa­lity to recei­ve highway inter­ven­ti­ons. Our iden­ti­fi­ca­ti­on stra­tegy allows us to iden­tify rele­vant bia­ses coming from both mea­su­re­ment error and omit­ted vari­a­bles. Our pre­fer­red esti­ma­tes point out a reli­a­ble road elas­ti­city in the ran­ge of 0.011 to 0.017. From this, we cal­cu­la­te a non­bi­a­sed return rate to highway infras­truc­tu­re of 21.3% in Bra­zil, pro­ving the high ren­ta­bi­lity of tho­se invest­ments in the deve­lo­ping world context.