TY - JOUR AU - Gordon, Roger AU - Li, Wei TI - Puzzling Tax Structures in Developing Countries: A Comparison of Two Alternative Explanations JF - National Bureau of Economic Research Working Paper Series VL - No. 11661 PY - 2005 Y2 - October 2005 DO - 10.3386/w11661 UR - http://www.nber.org/papers/w11661 L1 - http://www.nber.org/papers/w11661.pdf N1 - Author contact info: Roger H. Gordon Department of Economics 0508 University of California, San Diego 9500 Gilman Drive, Dept. 0508 La Jolla, CA 92093 Tel: 858/534-4828 Fax: 858/534-7040 E-Mail: rogordon@ucsd.edu Wei Li 1 East Chang An Avenue Oriental Plaza, Tower E1/10F CKGSB Beijing, China 100738 China Tel: +86-10-85188858 E-Mail: wli@ckgsb.edu.cn M1 - published as Roger Gordon, Wei Li. "Puzzling Tax Structures in Devloping Countries: A Comparison of Two Alternative Explanations," in Takatoshi Ito and Andrew K. Rose, editors, "Fiscal Policy and Management in East Asia" University of Chicago Press (2007) AB - Observed economic policies in developing countries differ sharply both from those observed among developed countries and from those forecast by existing models of optimal policies. For example, developing countries rely little on broad-based taxes, and make substantial use of tariffs and seignorage as nontax sources of revenue. The objective of this paper is to contrast the implications of two models designed to explain such anomalous policies. One approach, by Gordon-Li (2005), focuses on the greater difficulties faced in poor countries in monitoring taxable activity, and explores the best available policies given such difficulties. The other, building on Grossman-Helpman (1994), presumes that political-economy problems in developing countries are worse, leading to worse policy choices. The paper compares the contrasting theoretical implications of the two models with the data, and finds that the political-economy approach does poorly in reconciling many aspects of the data with the theory. In contrast, the forecasts from Gordon-Li model are largely consistent with the data currently available. ER -