Zambia VAT incentive experiments: version 2

All study firms are asked to request and retain receipts and invoices when making purchases from their suppliers. At each follow-up visit, treated firms are provided with the opportunity to play a lottery-style game to win a cash prize. The probability of winning cash is increasing in the total value of VAT invoices retained.

We randomize firms into three groups:

- Control -- no lottery or payout
- Low Treat -- lottery pays out 35 ZWK
- High Treat -- lottery pays out 300 ZWK

The lottery is a game of chance in which the respondent has the opportunity to draw a card from a hat. There are 10 cards: 1 yellow, 9 black. The respondent wins if they draw the yellow card. The number of draws (with replacement) per round is a function of the total value of VAT invoices retained by the respondent. The number of draws is assigned as follows:

0 ZWK in VAT invoices, 0 draws;
1-1500 ZWK in VAT invoices, 1 draw;
1500-3000 ZWK in VAT invoices, 2 draws;
3000+ ZWK in VAT invoices, 3 draws

Treated firms are also entered in a single lottery at the end of all follow-up visits, where their probability of winning is a function of their share in all VAT invoices collected during the entire study. This is to create some incentive to continue retaining invoices beyond the threshold of 3000 ZWK.

The VAT invoices that are eligible for the study are only those for the purchase of inventory or capital improvements for the enrolled retail firm. Invoices are only valid if they reflect purchases made since our most recent visit. We aim to visit each firm every 3-4 weeks.

Firms in all three groups were given information to help them recognize valid VAT invoices.

We are especially interested in estimating heterogeneous effects by two variables from the baseline survey: (i) the share of recent purchases made from regular suppliers, and (ii) an index of the strength of relational contracts with suppliers. The index will be based on whether the firms' regular suppliers (1) provide goods on credit, (2) offer price discounts, (3) give favorable delivery terms, (4) provide extra products as a gift for regular business, and (5) provide other goods or services not formalized in a contract.

With the above heterogeneity analysis, our goal is to test whether firm-to-firm relationships constrain tax policy. Specifically, we aim to these the hypotheses that when incentivized to demand VAT invoices from their suppliers, (i) retail firms with strong supplier relationships will be less likely to switch suppliers, and (ii) retail firms with strong supplier relationships will be less likely to demand VAT invoices from their existing suppliers. The idea underlying these tests is that tax avoidance may be an implicit part of the relational contract between firms, with benefits accruing to buyers in the form of lower prices and/or the provision of other services, conditional on them not requesting formal tax invoices for their purchases.

A note about the IRB approval for this study. We received Cornell IRB approval for the pilot study in 2021, which involved a similar incentive for households to request and retain VAT invoices. I am going to list that IRB on this application, because it covered the same interventions applied in the same setting, but to a different population. After the pilot we redesigned the study to focus on small firms rather than households. We determined that the experiment with firms is not human subjects research by definition. Our in-country partners at the Zambia Revenue Authority also concluded that no IRB review was needed for this study, and authorized the study to proceed under the ZRA's standing permission to conduct research in Zambia. The director of research at ZRA provided a letter of approval and introduction for the study.

2023-02-28

2023-07-31

1. Total value of VAT invoices received when making purchases from suppliers or capital improvements to the business
2. Share of stocking or capital improvement transactions that lead to the receipt of a valid VAT invoice (as a share of all stocking/capital improvement purchases)
3. VAT-related shopping behaviors, including switching to a VAT-registered supplier, reallocating business to an existing VAT-registered supplier, and inquiring after the VAT registration status of an existing or potential supplier
4. The share of stocking/capital purchases from regular suppliers, where regular suppliers are those from whom the firm has made at least two recent purchases, and from which it plans to purchase again
5. The strength of relational contracts with suppliers

We measure the strength of relational contracts with suppliers using an index based on whether the firms' regular suppliers (1) provide goods on credit, (2) offer price discounts, (3) give favorable delivery terms, (4) provide extra products as a gift for regular business, and (5) provide other goods or services not formalized in a contract. We also investigate these outcomes separately.

We randomize 1083 small retail firms in Lusaka, which we enrolled in a baseline survey, into three treatment arms:

Control (407 firms)
Low treatment (338 firms)
High treatment (338 firms)

Treated firms receive either a low or a high financial incentive to request and retain VAT invoices from their suppliers. Treatment assignment is at the individual firm level, stratified on key characteristics that we identified at baseline.

We stratified firm assignment on the following characteristics:

1. Percentage of recent stocking purchases made from regular suppliers (0, 1-50, 51-99, 100).
2. Above/below median share of recent stocking purchases for which a VAT invoice was received
3. Above/below median total recent stocking expenditure
4. Market size (categorical)

Intervention delivery occurred 2-3 months after completion of a baseline survey. Firms were identified for the baseline survey by systematically walking through commercial areas of Lusaka to find and approach small retail firms that are not registered for the VAT (and hence not already incentivized to request tax invoices from suppliers).

Stata.

Individual firm.

No

Treatment assignment is not clustered.

1083 firms, for a minimum of 4 follow-up surveys (after the baseline survey, and a brief survey conducted during intervention delivery).

Not a clustered RCT.