Why is license plate rationing not a good transport policy?

License plate rationing (LPR) has become an increasingly popular travel demand management strategy in developing countries. This paper aims to show that such a trend is concerning because of the inherent flaws of the policy. Using a conceptual mathematical model that explicitly considers mode choice and user heterogeneity, we prove that LPR is not a first-best policy, even though it can help reduce congestion under certain conditions. There is however no guarantee that LPR can even be a second-best policy: it may never improve the system cost, regardless of its configuration. More important, we show when it does improve the system cost, LPR makes it impossible to maximize the system efficiency by using other complementary policies. Our analysis also reveals unintended consequences of the policy. Notably, a more restrictive LPR could actually make congestion worse when imposed on a rich population that can bypass the policy through acquiring additional cars. Finally, numerical results confirm that LPR is likely regressive in many (if not most) practical cases. Specifically, the wealthier users tend to benefit more from LPR, especially when they can buy additional cars. The regressivity of the policy may be corrected only when LPR is highly restrictive and additional vehicle acquisition is tightly controlled. However, such a correction may only be achieved at the expense of significant welfare losses.