English Till intranät

The information method - theory and application

Författare: Per Engström, Och Patrik Hesselius, Och

Sammanfattning av Working paper 2007:17

When estimating the extent of e.g. excess use of public benefits one traditionally uses direct monitoring. Such direct estimates are afflicted with an intrinsic negative bias since you only count what you find. This paper presents and assesses an alternative intuitive, yet relatively unexplored, approachthatmay reduce thebiasbymaking use of the individual’s own response to information of increased monitoring. Through an extensive randomized social experiment we apply the method to one particular Swedish public benefit: Parental Benefit for Temporary Childcare. In our view the application was successful: the results are interpretable and we are able to surface more hidden excess use through the information method. As a rough estimate we find that the information based estimate of excess use is 40 percent higher than the corresponding estimate based on ordinary random monitoring (22.5 percent compared to 16 percent). The method is potentially applicable to a large number of related fields, such as e.g. tax evasion and insurance fraud.
JEL: C51, C93, H55. Keywords: Monitoring, Social insurance, Randomized experiments

When estimating the extent of e.g. excess use of public benefits one traditionally uses direct monitoring. Such direct estimates are afflicted with an intrinsic negative bias since you only count what you find. This paper presents and assesses an alternative intuitive, yet relatively unexplored, approach that may reduce the bias by making use of the individual’s own response to information of increased monitoring. Through an extensive randomized social experiment we apply the method to one particular Swedish public benefit: Parental Benefit for Temporary Childcare. In our view the application was successful: the results are interpretable and we are able to surface more hidden excess use through the information method. As a rough estimate we find that the information based estimate of excess use is 40 percent higher than the corresponding estimate based on ordinary random monitoring (22.5 percent compared to 16 percent). The method is potentially applicable to a large number of related fields, such as e.g. tax evasion and insurance fraud.

Keywords: Monitoring, Social insurance, Randomized experiments
JEL: C51, C93, H55.

Forskningsområden