Please fill out this short user survey of only 3 questions in order to help us improve the site. We appreciate your feedback!
Information, Fiscal Capacity and Tax Revenues: An Experimental Evaluation
Last registered on May 05, 2018


Trial Information
General Information
Information, Fiscal Capacity and Tax Revenues: An Experimental Evaluation
Initial registration date
May 04, 2018
Last updated
May 05, 2018 6:55 PM EDT
Primary Investigator
University of Warwick
Other Primary Investigator(s)
PI Affiliation
Columbia University
PI Affiliation
Paris School of Economics
PI Affiliation
Columbia University
Additional Trial Information
On going
Start date
End date
Secondary IDs
The value-added tax (VAT) is one of the most important sources of domestic tax revenue in developing countries. In theory, the VAT has two desirable properties: it is more efficient than alternative tax instruments, and it is self-enforcing because the two sides of each transaction (seller and buyer) have incentives to report truthfully. However, when enforcement capacity is limited, these two properties can break down, leading to economic distortions and tax revenue losses. When cross-checking the transactions reported monthly by all VAT-registered firms in Uganda in the period 2015-2017, we estimated substantial discrepancies leading to a large revenue shortfall.
This experiment aims to improve compliance along the VAT chain through the implementation of a tax enforcement intervention that encourages taxpayers to amend their past tax returns and comply with tax laws in the future. Specifically, the treatment will consist of sending notification letters to about 1,000 randomly selected firm-pairs (i.e., buyer-seller pairs) with the following information: (i) their VAT returns will be closely monitored (ii) a list of up to 3 discrepancies that have been found in their past returns (iii) failure to comply could result in additional enforcement measures or prosecution. We will measure the direct effect of the letters on the tax compliance behavior of the treated firms, and also the spillover effects on their trading partners.
External Link(s)
Registration Citation
Almunia, Miguel et al. 2018. "Information, Fiscal Capacity and Tax Revenues: An Experimental Evaluation." AEA RCT Registry. May 05. https://doi.org/10.1257/rct.2958-1.0.
Former Citation
Almunia, Miguel et al. 2018. "Information, Fiscal Capacity and Tax Revenues: An Experimental Evaluation." AEA RCT Registry. May 05. http://www.socialscienceregistry.org/trials/2958/history/29165.
Sponsors & Partners

There are documents in this trial unavailable to the public. Use the button below to request access to this information.

Request Information
Experimental Details
The intervention consists of sending notification letters to a randomly-selected sample of firm pairs (i.e., a seller-buyer match) for which reporting discrepancies were observed in the year before the intervention. The letters state that the tax authority has developed a new system to monitor VAT compliance and that the VAT returns will be closely monitored. The letters list up to 3 detected discrepancies and ask taxpayers to submit an amendment to correct any existing discrepancies. Finally, the letters point out that a failure to comply can result in additional enforcement measures or prosecution.
Intervention Start Date
Intervention End Date
Primary Outcomes
Primary Outcomes (end points)
Retrospective outcomes: amendment of past returns (binary), amount amended (in million UGX), change in pre-treatment VAT liability after amendment (in million UGX).

Forward-looking outcomes: actual vs. predicted VAT liability in post-treatment returns, probability that the VAT liability is positive, probability that VAT liability in a given month is higher than the VAT liability in the same month the year before, probability of misreporting, amount misreported, probability of underreporting, amount underreported, continuation of the trading relationship, reported transaction volume for the pair (in million UGX).
Primary Outcomes (explanation)
We define by misreporting a case where the amount reported by the seller (S) is smaller than the amount reported by the buyer (B), i.e. a case where S<B. Similarly, we define underreporting as a case where S>B.

Note: we will measure all outcomes both for firms receiving notification letters directly, as well as for their trading partners, whose behavior may be affected through spillover effects. We will also analyze spillover effects on other trading partners of the treated firms.
Secondary Outcomes
Secondary Outcomes (end points)
Secondary Outcomes (explanation)
Experimental Design
Experimental Design
The first step of the experimental design selects a "base sample" of firm-pairs. The criteria to be included in this base sample are: both firms in the buyer-seller pair must be registered for VAT, must have filed a VAT return in the last month, must not be part of the URA's annual audit plan for the financial year (2017/18), and must have an accumulated discrepancy above 1 million UGX (approximately 270 USD) over the previous 10 months.

In the second step, we employ an iterative randomization procedure to select firm-pairs to be included in the "study sample". Once the first firm-pair is randomly selected, we remove all the trading partners of the two firms (buyer and seller) from the base sample. Then, we randomly select another firm-pair and remove all their trading partners. Following this iterative procedure, we obtain a set of firm-pairs which are (i) separated by at least two nodes in the network of firm-to-firm transactions and (ii) have an accumulated misreporting discrepancy that is greater than 270 USD over the past year.

In the third step, we apply a stratified randomization procedure to allocate firms from the study sample to three treatment arms and a control group. In the first treatment arm, the letter is sent only to the buyer in the selected pair. In the second treatment arm, the letter is sent to the seller. Finally, in the third arm, letters are sent to both the buyer and seller.
Experimental Design Details
The criteria for the first step were determined in collaboration with the Uganda Revenue Authority. Our base sample consists of 15,983 firm-pairs. In the second step, we obtain a study sample of 1,673 firm-pairs. In the third step, we randomly assign 40% of firms (677) to be part of the control group. The remaining 60% (996) are divided up equally across the three treatment arms. In this randomization process, we stratify by the ratio of sales to final consumers, firm size (measured by total revenue from sales), and by whether or not the firms' headquarters are located in or outside Kampala (the capital city of Uganda).
Randomization Method
The randomization was conducted in office using a computer.
Randomization Unit
Firm-pair, defined as a buyer-seller pairs that trade with each other in the year prior to the intervention.
Was the treatment clustered?
Experiment Characteristics
Sample size: planned number of clusters
1,673 firm-pairs.
Sample size: planned number of observations
1,673 firm-pairs (3,346 firms).
Sample size (or number of clusters) by treatment arms
(1) a letter is sent to the buyer – 335 firm-pairs.
(2) a letter is sent to the seller – 332 firm-pairs.
(3) a letter is sent to the buyer and the seller – 329 firm-pairs.
(4) no letter was sent (control group) – 677 firm-pairs.
Minimum detectable effect size for main outcomes (accounting for sample design and clustering)
IRB Name
Columbia University's Institutional Review Board
IRB Approval Date
IRB Approval Number
Post Trial Information
Study Withdrawal
Is the intervention completed?
Is data collection complete?
Data Publication
Data Publication
Is public data available?
Program Files
Program Files
Reports, Papers & Other Materials
Relevant Paper(s)