Does Hypercongestion Exist? New Evidence Suggests Not

Transportation engineers are taught that as demand for travel goes up, this decreases not only speed but also the capacity of the road system, a phenomenon known as hypercongestion. We revisit this idea. There is no question that road systems experience periods in which capacity falls. However, we point out that capacity is determined by both demand and supply. Road construction, lane closures, stalled vehicles, weather, and other supply shocks provide an alternative explanation for the empirical evidence on hypercongestion. Using data from the Caldecott Tunnel in Oakland, California, we show that a naive regression recovers the standard hypercongestion result in the literature. However, once we use instrumental variables to isolate the effect of travel demand this effect disappears and across specifications we find no evidence that capacity decreases during periods of high demand. This lack of evidence of hypercongestion calls into question long-standing conventional wisdom held by transportation engineers and implies that efficient "Pigouvian" congestion taxes should be substantially lower than implied by hypercongestion models