28
Appendix 2. Main Elements of a VAT Refund Verification Program
An effective refund verification program should provide for (1) an initial risk assessment of new VAT
applicants, (2) a front-end data validation system, (3) risk-based audit selection, and (4) differentiated
verification methods.
Initial risk assessment. Refund management systems should include an initial risk assessment to
identify potentially fraudulent registrants. When the taxpayer first registers with the tax
administration, the registration application should be compared with standardized risk profiles.
Suspect applications should be referred for a preregistration visit by a tax officer to determine whether
the location and nature of the business are consistent with the declared business activities, sources of
supplies, and financing. Applications that remain suspect after the visit should be denied registration
or conditions should be established for registration (e.g., posting of a security).
Front-end data validation. Refund claims should first be subjected to a front-end check of
completeness (including supporting documentation), arithmetic correctness, and consistency of
information before being screened for audit selection. The tax agency should have the authority to
override minor arithmetic errors; more serious errors should lead to a rejection of the claim and
require its resubmission by the taxpayer.
Risk-based audit selection. Once validated, refund claims should be classified into different risk-
based categories ( e.g., high, moderate, low) for audit selection. The criteria for establishing the
rankings could include such factors as the claim’s deviation from amounts claimed previously, the
claimant’s compliance history, whether the refund represents a first-time claim, and the date of the
most recent audit. The amount of audit resources allocated to refund verifications should be consistent
with the degree of revenue risk posed by refund claims relative to other compliance risks faced by the
tax agency. From this perspective, an effective audit program would devote, for example, 30 percent
of its audit resources to verifying refund claims and 70 percent of resources to other types of audits.
Differentiated verification methods. Different types of verifications should be applied, depending
on a claim’s degree of risk. High-risk claims should be audited prior to issuing the refund, usually at
the taxpayer’s premises. Moderate-risk claims would be assigned an upper-refund limit. Claims not
exceeding the upper limit and accompanied by supporting documents would be subject to a brief desk
review before the refund is issued and a post-approval audit at periodic intervals. Claims exceeding
the limit would be subject to a pre-approval audit. Low-risk refund claims would typically be
approved within 72 hours without the pre-approval audit requirement. For these claims, a post-
approval audit may be arranged on a random basis from time to time.
When an audit is to be conducted, the scope of the examination should be limited to the refund claim
(and not cover other tax types) and focus on issues material to the refund. Statistical sampling
methods should be used to determine the number of transactions that need to be checked. Importantly,
the audit should not rely excessively on export documents from Customs, since such documents are
not really evidence that the goods in question were in fact exported. Instead, a sample number of
export shipments should be selected and traced through purchase invoices, sales records, inventory
records, shipping receipts, and payment and bank records. This would be better proof of the validity
of the exports. And, if these records still leave doubt, the auditor may confirm with Customs of the
importing country that the goods were properly entered for consumption.