A more accurate measurement of tax effort

Abstract

Even though it is commonplace in many countries for federal governments to give fiscal support to local governments, this may work as a deterrent in providing incentives to local governments to increase their own revenue. To address this problem of moral hazard, various techniques of measuring the local governments’ tax effort have been suggested in the literature. However, all of these techniques share a drawback: ‘biasedness’. To solve this problem, this article presents an unbiased estimator, which is the Kalman filter estimator. A Monte Carlo simulation shows evidence that the Kalman filter estimator is more accurate than other methods suggested in the literature.

Acknowledgements

I am grateful to the seminar participants at the 60th Congress of the International Institute of Public Finance held at Milan in 2004 for their constructive comments. However, all errors remain the author's.

Notes

¹ This assumption serves the purpose of exposition. The same logic applies to more complicated processes.

² For the derivation of Kalman filter equations, see Burmeister et al . (1982).

³ That is,

⁴ The reason for using a single equation instead of the state-space equations such as Equations 4 and 5 for the data generation is not to treat the Kalman filter estimation more favourably than the others. If the data are generated with Equations 4 and 5, the Kalman filter estimator results in more accurate estimates for tax effort.