DRA-1-06/2-01-CO:R:C:E 224295 PH

Regional Commissioner of Customs
North Central Region
55 East Monroe Street
Chicago, Illinois 60603-5790
ATTN: Chief Region Drawback Branch

RE: Protest 3901-92-100583; Substitution Manufacturing Drawback; Customs Audit; Random Sample; Statistical Analysis Dear Sir:

The above-referenced protest was forwarded to this office for further review. We have considered the points raised by your office, the protestant, and the materials in the file. Our decision follows.

FACTS:

The protest is of the liquidation of 57 drawback entries (or claims) dated August 7, 1986, through October 26, 1989. The protestant had received $807,260 in accelerated payments of drawback for the entries under protest. The entries were the subject of a Customs audit (Report Control No. 321-89-DRO-006, dated August 6, 1991). During the course of the audit, personnel of the protestant met with Customs officials. The protestant submitted amended entries (of the 57 drawback entries under consideration) on May 15, 1990, August 27, 1990, and December 4, 1990. The protestant conceded a number of errors in its original entries and returned $203,294 (received by Customs on October 1, 1990) of the $807,260 in accelerated payments.

Using established statistical analysis procedures, (i.e., establishing the applicable universe, determining the acceptable rate of error, and using the appropriate tables for sampling size and random selection of samples), a randomly selected sample of the designated imported parts upon which drawback was claimed in the 57 entries under consideration was selected. This sample selection was applied to the second amendment (that submitted on August 27, 1990) because, according to the audit report, the sample had already been agreed upon when the third amendment was filed. According to the Customs Regional office which performed the audit, a list of the 291 parts selected to be sampled was provided to the protestant and the protestant agreed and approved the selected parts on November 20, 1990. The protestant also states that it agreed to the sample (although after the protestant reviewed the final results it contended that the sample was distorted).

The above sampling procedures resulted in the audit of a sample of 291 parts out of a universe of 6,580 parts in the entries under consideration. Of the 291 parts sampled, deficiencies were found with regard to 120 parts. The total duty refunded as drawback for the 291 sampled parts was $40,371 ($138.732 per part). The total duty refund which the audit report stated should have been disallowed was $18,442 ($63.375 per part). This disallowance per part was projected to the total universe of parts, resulting in a calculation that $412,837 should be disallowed (i.e., $63.375 X 6,580 parts = $417,007 X 99% = $412,837).

The entries were liquidated in accordance with the above advice between December 20, 1991, and February 7, 1992. The liquidations of the entries were protested on March 17, 1992. In addition to contending that the sample selected for audit distorted the error rate for the entries, the protestant specifically discussed certain items cited in the audit report. These issues are analyzed in the LAW AND ANALYSIS portion of this ruling.

ISSUE:

Is there authority to grant the protest of the denial of drawback described in the FACTS portion of this ruling?

LAW AND ANALYSIS:

Initially, we note that the protest, with application for further review, was timely filed under the statutory and regulatory provisions for protests (see 19 U.S.C. 1514 and 19 CFR Part 174). We note that the refusal to pay a claim for drawback is a protestable issue (see 19 U.S.C. 1514(a)(6)).

The drawback entries which are protested are for drawback under 19 U.S.C. 1313(b). Basically, section 1313(b), often called the substitution manufacturing drawback law, provides that if imported duty-paid merchandise and duty-free or domestic mer- chandise of the same kind and quality are used in the manufacture or production of articles within three years of the receipt of the imported merchandise by the manufacturer or producer of the articles and articles manufactured or produced from the imported duty-paid merchandise or duty-free or domestic merchandise are exported or destroyed within five years of the importation of the duty-paid merchandise, 99 percent of the duties on the imported duty-paid merchandise shall be refunded as drawback, even if none of the imported merchandise was actually used in the manufacture or production of the exported articles. The Customs Regulations pertaining to drawback are found in 19 CFR Part 191. The protestant claimed drawback under the general drawback rate for articles manufactured with the use of component parts (published in Treasury Decision (T.D.) 81-300).

The specific findings in the audit report addressed in the protest are discussed below.

On page 7 [page references are to the audit report], paragraph a) [at top of page], the report states that quantities of the designated imports were greater than the quantities actually imported. The protestant concedes this error.

On page 7, paragraph b), the report states that a non- existent consumption entry was used to designate imported merchandise. The protestant appears to concede this error (i.e., the error is addressed in the "corrections" section of the protest and is stated to represent a misreading of the consumption entry number). The protest is DENIED in this regard.

On page 7, paragraph c), the report states that the duty claimed per part was greater than the duty actually paid. The protestant concedes this error.

On page 7, paragraph d), the report states that the imports designated were duty-free merchandise, although drawback was claimed. The protestant concedes this error.

On page 7, paragraph e), the report states that the protestant was unable to relate Certificates of Delivery against vendors' invoices and inventory records. The protestant attempts to explain this deficiency by stating that import vendors supplied the protestant with certificates of delivery showing delivered quantities which often would not match quantities received by the protestant. In this regard, we note that the Customs Regulations require that a Certificate of Delivery in this situation "must describe the merchandise delivered, tracing it from the custody of the importer to the custody of the manufacturer" (19 CFR 191.65(a)). The protest is DENIED in this regard.

On page 7, paragraph f), the report states that the protestant did not provide records for a number of specified reasons. The protestant contends that this statement is not supported in the audit report by specific references and should be ignored. The material which we have available to us in the file does not indicate which parts sampled have this deficiency. To the extent that drawback is denied on this ground without reference to a specific sampled part and specific records, the protest is GRANTED in this regard (otherwise, the protestant is denied the right of protest for that item). However, if drawback is denied on this ground with regard to a specific sampled part and specific records, the protest is DENIED in this regard for the following reasons. The Customs Regulations (19 CFR 191.32, 191.2(o), 191.5, 191.10) require the keeping of certain records for drawback and that the records be available for examination by Customs. In its drawback contract, the protestant agreed to maintain certain described records and to make the records available for audit by Customs. The protestant stated it understood that "drawback is not payable without proof of compliance" and that "if [its] records do not show that [it] satisf[ies] those legal requirements [relating in part to record-keeping], drawback cannot be paid."

On page 7, paragraph a) [bottom of page], the report states that certain designated imports were not used in production and were in inventory. The protestant concedes that its records did not correctly show the parts going into the factory but states it could not build the articles without this "single-sourced" part. Therefore, the protestant contends, this error really had no effect on drawback. The protestant also appears to contend that it could have established compliance with this requirement (i.e., timely use of the imported designated merchandise) by use of a first-in, first-out (FIFO) accounting method. With regard to the lack of proper records, see discussion of record- keeping requirements, above. In the drawback contract signed by the protestant, the protestant specifically agreed to keep records to establish compliance with this requirement. As to the protestant's arguments on the use of the FIFO accounting method to establish the timely use of the imported designated merchandise, it is true that "if FIFO accounting records are used and these records prove use for drawback purposes, these records can be the basis for allowance of drawback" (C.S.D. 79-301). However, there is no evidence in the file to show that the conditions precedent for use of the FIFO accounting method in the above quotation are met. The protest is DENIED in regard to this issue.

On page 8, paragraph b) [top of page], the report states that designated imported parts were identified on the drawback claims as two different part numbers, although they were one and the same part, resulting in the protestant claiming drawback twice on the same designated imports. The report refers to Exhibit 2 in this regard. The protestant does not comment on this item. The parts referred to in Exhibit 2 were not, themselves, part of the statistical sample, although the Exhibit does illustrate unrebutted (by the claimant) deficiencies in the claims. Because this ruling is based on the sampled parts, the deficiencies shown by Exhibit 2 do not affect the protest decision.

On page 8, paragraph c) [top of page], the report states that designated imported parts were rejected as not meeting the protestant's quality control standards. The protestant states that it was having a problem with a supplier and there was a lot of movement in and out of the factory with this part. The protestant states that this "single-sourced" part was necessary for the manufactured article and, therefore, this error really had no effect on drawback. See discussion of paragraph a), page 7 [bottom of page], in this regard. The protest is DENIED, on the same basis, in regard to this issue.

On page 8, paragraph d), the report states that the protestant was unable to relate Certificates of Delivery against vendors' and certain worksheets and reports. See, in this regard, discussion of paragraph e), page 7. The protest is DENIED, on the same basis, in regard to this issue.

On page 8, paragraph e), the report states that the protestant did not provide records for a number of specified reasons. See, in this regard, discussion of paragraph f), page 7. On the same basis, to the extent that drawback is denied on this ground without reference to a specific sampled part and specific records, the protest is GRANTED in this regard and, if drawback is denied on this ground with regard to a specific sampled part and specific records, the protest is DENIED in this regard.

On page 8, paragraph a), the report states that designated imported parts claimed were never used to produce the exported model (citing Exhibit 3). The protestant does not directly address this item, but questions the apparent indication by the report that designated imported parts must be used to produce the exported model (in drawback under 19 U.S.C. 1313(b), of course, although the designated imported merchandise and the substituted duty-free or domestic merchandise must be used in a manufacture or production, either or any combination of both may be used in the manufacture or production of the exported article). We conferred with the appropriate local Customs official in this regard and found that the deficiency in this item is that no part of the same kind and quality as the designated imported part appeared in the exported article. We note that each of the parts listed in Exhibit 3 is one of the parts selected in the sample for audit. The protest is DENIED in this regard.

On page 8, paragraph b) [bottom of page], the report states that discontinued, obsolete and cancelled designated imports were claimed to have been used to produce the exported models (citing Exhibit 4). See, in this regard, discussion of paragraph b, page 8 [bottom of page]. The protest is DENIED, on the same basis, in regard to this issue.

On page 8, paragraph c) [bottom of page], the report states that quantities of parts claimed on drawback entries exceeded the "bill of materials quantities" (citing Exhibit 5; this exhibit clarifies this item, describing the deficiency to be one in which the quantity of parts actually appearing in the manufactured article, according to the bill of materials, is less than the quantity of the same parts claimed to appear in the manufactured article). The protestant contends that this statement is not supported in the audit report by specific references and should be ignored. As explained above, this item does appear to be supported in the audit report by specific references (see Exhibit 5); clearly an overpayment of drawback results if more parts are claimed for drawback than appear in the manufactured article which is exported. Each of the parts listed in Exhibit 5 appears to be one of the parts selected in the sample for audit, with the possible exception of part AT 38206. The protest is DENIED in this regard, except with regard to part AT 38206 (if it is not one of the parts selected in the sample for audit, in which case the protest is GRANTED in this regard, for this part).

On page 9, paragraph a), the report states that quantities of models claimed on the drawback entries to have been exported exceeded the quantities of models actually shipped and exported per the exportation records examined (citing Exhibit 6). The protestant does not comment on this item. A drawback claimant is specifically required to establish exportation of articles for drawback purposes by the Customs Regulations (19 CFR 191.51 et seq.) In the drawback contract signed by the protestant, the protestant specifically stated that it understood that "it must be established that the completed articles were exported within 5 years after the importation of the imported merchandise." The protest is DENIED in this regard.

On page 9, paragraph b), the report states that exported models on which drawback was claimed were shipped to Puerto Rico and, therefore, drawback should not have been allowed on those models. The protestant concedes this error.

On page 9, paragraph c), the report states that certain exports were claimed, for drawback purposes, more than once (citing Exhibits 6 and 7). The protestant does not comment on this item. See discussion of paragraph a), page 9, in this regard. The protest is DENIED in this regard.

On page 9, paragraph d), the report states that the protestant was unable to provide exportation records on shipments between November 1, 1983, and April 28, 1987. The protestant contends that it could have provided records (i.e., invoice records invoicing a foreign location and a payment record for the same), but it was unable to provide records in the format requested by Customs. The protestant contends that such records should be satisfactory. As stated in our comments on paragraph a), page 9, the Customs Regulations have specific requirements for the establishment of exportation for drawback purposes and the protestant agreed to establish exportation in its drawback contract. The cited Customs Regulations provide several alternative kinds of evidence to establish exportation. Unless the protestant meets one of these alternative requirements, the protest is DENIED in this regard.

In addition to the specific items in the audit report discussed above, the protestant comments on certain more general items in the audit report (see pages 5 and 6 of the report). These more general items do not appear to have had any effect on the specific recommendations made in the audit report (i.e., the conclusion of the audit report is that there should be a disallowance of drawback for: (1) "receipt and used-in production of the designated imports"; (2) "used-in production of the designated imports to produce the exports" [as noted in the discussion of paragraph a), page 8, this deficiency is actually that no part of the same kind and quality as the designated imported part appeared in the exported article]; and (3) "shipment of exports"; and these three groups are dealt with in the specific items discussed above). Therefore, we are not commenting on the more general items, except insofar as our comments on the specific items relate to them.

The protestant also contends that the statistical sample used in this case highly distorted the error rate applied to the drawback entries. In support of this contention, the protestant notes that the average drawback claimed for the sampled parts was $138.73 although the average drawback claim for the universe of 6,590 parts was $92.91. The protestant states that it understood the purpose of the use of the statistical sample in this case to be to determine a mean rate of error, with a maximum tolerable rate of error of 10%. Noting that the difference in claimed drawback in the sampled parts and in the entire universe is $45.82 per part, or 49.3%, the protestant contends that this illustrates the distortion in the sample.

The Courts have approved the use of statistical analysis in various situations (see, e.g., Castaneda v. Partida, 430 U.S. 482, 97 S.Ct. 1272 (1977), and cases cited therein; see also Texpor Traders, Inc. v. Trust Co. Bank, 720 F. Supp. 1100 (S.D. N.Y. 1989), in which the Court took judicial notice, "that in statistical analysis, using a well known mathematical theorem, viz, the Central Limit Theorem, a sample size of thirty or more is generally recognized as sufficient to guarantee normality of the distribution of sample means" (Note 2, 720 F. Supp. at 1105); and Bright, Kadane, and Nagin, Statistical Sampling in Tax Audits, 13 JOURNAL OF THE AMERICAN BAR FOUNDATION, Law & Social Inquiry 305 (1988), see in particular pp. 310-318).

We also note that the Congressional Committees with oversight of the drawback laws have recognized the validity of the use of sampling as a drawback audit technique. Public Law 103-182, the "North American Free Trade Agreement Implementation Act", has been enacted by both Houses of Congress and was signed into law by the President on December 8, 1993. Section 632 of Public Law 103-182 contains a number of amendments to the drawback law (19 U.S.C. 1313). In the House and Senate reports on H.R. 3450, the bill which was enacted as Public Law 103-182, it was stated in regard to drawback that:

... [T]he Committee expects that, if the entire universe of the claimed import entries and exports is audited, and the audit reveals that only a portion of a company's claims are deficient, drawback should be denied only on that portion found to be deficient. However, if only a representative sample of the claimed import entries and exports is audited, and the audit reveals that a significant portion of the audited claims is deficient, then denial of the audited company's drawback claims may extend beyond the portion audited. [H. Report 103-361, 103d Cong., 1st Sess., 132 (1993); see also, S. Report 103-189, 103d Cong., 1st Sess., 84 (1993), which contains similar language.]

In this case the sample was selected using established statistical analysis procedures (the applicable universe was established, an acceptable rate of error was determined, and the appropriate tables for sampling size and random selection of samples were used). After selection of the parts to be audited in the sample, the protestant was given a list of the selected parts and approved the sample selected. To this point in the process, we see no inconsistency with the above authorities in the sampling process. However, according to both the protestant and the audit report, "a maximum tolerable rate of error of 10%" was agreed upon. According to the calculations in the audit report, the rate of error was more than 67% (of the $611,369 stated to be included in the amended claim, $412,832 was disallowed), far in excess of the agreed upon maximum tolerable rate of error. (Even with the correction to the calculations in the audit report discussed below, the rate of error was much more than that which was agreed to be acceptable.)

Upon finding an unacceptable rate of error, Customs should have either given the claimant an opportunity to correct its claims and re-audited the corrected claims with the use of an agreed-upon sample or performed a 100% audit (i.e., without recourse to statistical sampling) and liquidated the claims accordingly. Instead, Customs proceeded to liquidate the claims, projecting the rate of error it calculated from the audit to the entire universe of drawback claimed in the entries.

Once liquidated, a liquidation may not be "unliquidated" (see United States v. Utex International Inc., 6 Fed. Cir. (T) 166, 857 F. 2d 1408 (1988)). Absent probable cause to believe there is fraud (see 19 U.S.C. 1521), the only action Customs is authorized to take on a liquidation which has been timely protested and for which the 90 days for voluntary reliquidation have elapsed (see 19 U.S.C. 1501) is to allow or deny the protest in whole or in part (see 19 U.S.C. 1515(a) (in this regard, see also 19 U.S.C. 1515(c) and (d), as added to section 1515 by section 617 of Public Law 103-182)).

Therefore, we must consider this protest as presented (i.e., even though Customs liquidated the entries on the basis of rates of error which were agreed upon by Customs and the protestant to be unacceptable). We conclude that the audit report erroneously projected the audit findings to the universe of drawback claimed in the entries. As stated above, 291 parts were sampled and 120 of those parts were found to have deficiencies. Thus, the simple rate of error was 41.24%. The total drawback claimed for the 291 parts sampled was $40,371 and the total drawback disallowed was $18,442. Thus, the rate of error based on drawback claimed for the sampled parts was 45.68% As stated above, either of these rates of error were agreed upon in advance to be unacceptable.

The audit report compounded the error of using an unacceptable rate of error by its calculations to determine the drawback which should be disallowed. This was done by using a multiple-step process (which decreases the level of reliability) in which the total drawback disallowed on the sampled parts ($18,442) was divided by the parts sampled (291) and the result ($63.375) was projected to the total parts in the universe. As stated above, this results in an effective error rate of in excess of 67% (of the $611,369 stated to be included in the amended claim, $412,832 was disallowed). The fallacy in the method used in the audit report is illustrated by projecting the total drawback (allowed and disallowed) per sample audited to the entire universe (i.e., $138.55 X 6580 = $948,999). This is approximately 55% more than the actual entire universe (stated to be $611,369).

We understand that the purpose of the projection method used in this case may have been to take into consideration the great variances in drawback per part. Such a concern is proper, but should be acted upon in the selection of the sample (e.g., by stratified sampling) (see, generally, in this regard the article by Bright, Kadane, and Nagin, cited above). As stated above, attempting to redress this concern in the projection of the rate of error, as was done in the audit report, decreases the level of reliability of the sample. We conclude that in this case, in which a simple random sampling method was used, the disallowance of drawback from the sample audited should have been projected on the basis of the simple rate of error (41.24%). The protest is GRANTED, in part, in this regard.

According to the audit report, the amount of drawback claimed on the amended entries (i.e., the August 27, 1990, amendments) was $611,369. The simple rate of error (41.24%) should be projected to this figure, so that the total disallowance of drawback is $252,129. Each of the protested drawback claims should be reliquidated using this simple rate of error.

[In its letter of September 20, 1993, in response to our letter of July 13, 1993, requesting the protestant to provide more information regarding this matter, the protestant again suggested that the sample selection process used in this case was flawed. The protestant alleges that the auditor knew that there were deficiencies with certain parts and included those parts in the sample, thus skewing the results. We disagree with this allegation. According to the evidence available to us, the sample was selected on a strictly random basis from the parts in the claims then before Customs (and before the December 4, 1990, amended claims were filed) and the protestant agreed to the parts selected (also before the December 4, 1990, amended claims were filed). Removing from the sample parts which, after the audit, were found to have deficiencies negates the purpose of the sampling process. Our decision in this matter is as stated below.] HOLDING:

The disposition of the Protest of denial of drawback described in this ruling is as follows:

The protest is DENIED, in part, and GRANTED, in part. The entries should be reliquidated in accordance with this decision. In accordance with Section 3A(11)(b) of Customs Directive 099 3550-065, dated August 4, 1993, Subject: Revised Protest Directive, this decision should be mailed by your office, with the Customs Form 19, to the protestant no later than 60 days from the date of this letter. Any reliquidation of the entry in accordance with the decision must be accomplished prior to mailing of the decision. Sixty days from the date of the decision the Office of Regulations and Rulings will take steps to make the decision available to Customs personnel via the Customs Rulings Module in ACS and the public via the Diskette Subscription Service, Lexis, Freedom of Information Act, and other public access channels.

Sincerely,

John Durant, Director
Commercial Rulings Division