New research noted
The logic of deliberately ineffective migration policy
"The Political Economy of Migration Enforcement: Domestic versus Border Control," Centro Studia Luca D'Agliano, Working paper No. 371
This working paper by Giovanni Facchini and Cecilia Testa (August, 2014), recently listed on eh-net, offers a plausible explanation for instances of longstanding ineffective (e.g., "broken") immigration policies in democratic countries, where one might otherwise expect electoral pressures for efficiency:
Voters often prefer immigration levels lower than what would be optimal for the overall economy. Some politicians ("populists") are eager to support such public preferences, others
("utilitarians") are reluctant. If voters lack clear information on how much money is spent on enforcing immigration policy, and how effectively it is spent, then "utilitarian" politicians can
set "restrictive targets" to please voters, but "relax enforcement" so that more foreigners are actually admitted.
If the actual volume of immigration is "large" [not clear if "large" means relative to the economy or to average voter preferences or both], then relatively ineffective policy tools [e.g. border controls] would likely be favored by a "utilitarian"-controlled government over relatively effective tools ["domestic" enforcement]. This result occurs because the ineffective methods, though more expensive relative to what they achieve, enable immigration which overshoots the target more significantly, thus getting closer to the economic optimum.
This situation can persist as long as voters lack clear information on the cost of immigration policy enforcement because "utilitarian" politicians appear to voters as taking "populist" positions, although actually enforcing them relatively inefficiently.
If actual immigration volume is "small," then it is cheaper for "utilitarians" to focus policy more on the relatively effective approaches (domestic enforcement) but underfund them so as to (also in this case) enable actual volumes above the target and closer to the optimal level.