Date Available


Year of Publication


Degree Name

Doctor of Philosophy (PhD)

Document Type

Doctoral Dissertation


Graduate School


Public Policy and Administration

First Advisor

Dr. David R. Agrawal


Consumption taxes play a major role in revenue generation for many developing countries. They are also used to encourage or discourage certain consumption behaviors. However, the administrative aspects of implementing these taxes can create challenges for the tax authorities, and lead to distortions in taxpayer behavior. To overcome these challenges, efforts have been made to simplify policies and utilize technology to improve administration and compliance. My dissertation aims to provide empirical evidence on the effects of three consumption or commodity tax policy reforms in Indonesia. The dissertation consists of three essays as follows.

In my first essay, titled “The effect of modernizing the self-enforcing “paper” trail in Value Added Tax”, I study a modernization policy under Indonesia's Value Added Tax (VAT) regime, where the government switched from a paper-based VAT invoice system to an online electronic system. To measure how the digitalization of the self-enforcing “paper” trail affects firms' behavior and revenue mobilization, I exploit a two-stage implementation that mandated the new electronic invoice system to start in July 2015 for taxpayers in 7 provinces (similar to states) and then in the rest of 27 provinces one year later. This policy creates a quasi-experimental variation that allows me to use the two-stage introduction in a generalized differences-in-differences estimation following Malkova (2018). I utilize the Indonesian tax administrative data from the monthly VAT returns filed from 2014 to 2017. I use the firms in the 7 provinces that adopted the electronic invoices early as the treatment group and the rest of firms in the 27 provinces that adopted the electronic invoices later. I find that the electronic invoices system leads to an immediate increase in taxable sales but no effect on taxable inputs, and it further translates to increased VAT liability. The impact is driven by firms’ sizes, particularly smaller firms. The effect of electronic invoices are also heterogenous under different institutional environment (tax office type) and firm’s sectors.

The second essay in my dissertation, titled “Toward fewer tax audit: Evidence from Indonesia“, utilizes another VAT administrative policy reform. The existing empirical literature has primarily focused on examining the impact of a higher probability of detection or audit in a tax system. However, we have little empirical evidence on the opposite situation when the probability of detection or audit gets lower. This essay fills the gap by studying firms' behavioral response when the tax authority in Indonesia removes a mandatory audit requirement for firms claiming a refund on VAT excess payments below IDR 1 billion (around $70,000). I use a difference-in-differences estimation by exploiting the variation in the treatment at the sector level to estimate the impact of this reform. I use firms in the sectors concentrating on export activity as the treatment group and firms in the sector less concentrated on export as the control group. For this essay, I also use the Indonesian tax administrative data from the monthly VAT returns filed from 2016 to 2020. The result show that the fast-track refund led to an 11.66 percent increased taxable inputs (partially significant), but no impact on productivity (export and domestic sales) and VAT liability. Consistent with the standard tax compliance model, this increases in taxable inputs indicates that fewer audit activity may facilitate tax evasion.

The third essay, titled “Tax stimulus during the COVID-19 pandemic: Evidence from Indonesia” investigates the impact of a commodity tax stimulus program on new car purchases introduced in response to the COVID-19 economic downturn. This program provides a luxury sales tax (LST) exemption for new car purchases where the eligibility requirement is based on engine sizes and domestic content ratio. My empirical design exploits comparable vehicles similar in price, type, market segment, engine sizes, and engine power but with variations in their eligibility for the program. Using event study and multi period difference-in-differences, I analyze the shift in sales number between the incentivized cars and their comparable before and during the COVID-19 pandemic. I use national-wholesale level data of monthly new vehicle purchases from January 2019 to December 2021. I find that tax stimulus increases purchases on eligible cars by 80 percent but only statistically significant at 10 percent confidence interval, which may be due to the small number of observations in my data. This translates to approximately 40 thousand additional units of eligible vehicles being purchase from April 2021 to December 2021.

Digital Object Identifier (DOI)

Funding Information

My doctoral study was supported by the Indonesia Endowment Fund for Education (LPDP) with scholarship recipient number 201906221914531, from August 2019 until June 2023.