Iii.a Supporting Statement Part A

III.A SUPPORTING STATEMENT PART A.docx

Annual Retail Trade Survey

OMB: 0607-0013

Document [docx]
Download: docx | pdf

Page | 9


OMB INFORMATION COLLECTION

Department of Commerce

U.S. Census Bureau

Annual Retail Trade Survey

OMB Control No. 0607-0013


PART A. JUSTIFICATION


1. Necessity of Information Collection


The Annual Retail Trade Survey (ARTS) covers employer firms with establishments located in the United States and classified in the retail trade sector as defined by the North American Industry Classification System (NAICS). The survey requests firms to provide annual sales, sales tax, e-commerce sales, year-end inventories, total operating expenses, purchases, and accounts receivable. We also request, for selected industries, sales and e-commerce sales by merchandise line.


The data collected in the Annual Retail Trade Survey provide a current statistical picture of the retail portion of consumer activity. These data are collected to provide a sound statistical basis for the formulation of policy by various government agencies, as well as to serve as a benchmark for the estimates compiled from the Monthly Retail Trade Report. Results will be made available, at the United States summary level, for selected retail trade industries approximately fourteen months after the end of the reference year.


As requested by the Bureau of Economic Analysis (BEA), every five years, in survey years ending in “2” and “7”, ARTS requests data on detailed operating expenses from firms. The last time ARTS collected detailed operating expenses was in 2018 for the 2017 survey year. The plan is to reinstate some of these questions in 2023 as part of the 2022 survey year ARTS data collection. For survey year 2020 (collected in 2021), the ARTS will also include improved language and ordering of the questions

on the value of inventories and inventory valuation method to ensure a better understanding and response to the questions from respondents. Effective with survey year 2020 and consistent with the agency’s goal of harmonizing content across all annual surveys as recommended by the National Academy of Sciences, ARTS will no longer collect accounts receivable data. Survey year 2019 estimates (to be released in February 2021) will be the final year that accounts receivable data are available to the public. (Details on the National Academy of Sciences recommendations are found in Item 8. Consultations Outside the Agency)


The Census Bureau published a pre-submission notice in the Federal Register on Monday, July 6, 2020 (Vol. 85, No. 129). The notice, which was located on pages 40199 and 40200, proposed additional questions on the ARTS related to the impact of the coronavirus pandemic on firms for survey year 2020. After internal discussions, the Census Bureau decided it will not include additional questions on this survey related to the impact that the coronavirus pandemic had on firms.


Every 5 years, ARTS requests data on detailed operating expenses; the last collection was in 2018 for the 2017 survey year. During the 2022 collection year that will occur in 2023, ARTS will again collect detailed operating expenses. ARTS estimates are publicly released based on the North American Industry Classification System (NAICS), which has been widely adopted throughout both the public and private sectors.


The ARTS introduced a new sample for survey year 2016. In order to link the old sample and new sample estimates, two years of data (2015 and 2016) were collected during the 2016 survey year. Two years of data are only requested during the first year of a new sample. For survey year 2020, the ARTS will continue to use the sample introduced for survey year 2016. Thus, only one year of data will be collected. Respondents will be asked to provide one year of data for all subsequent years until a new sample is selected. A new sample will not be selected during this three-year clearance window.


This request is for the clearance of two electronic worksheets, the SA-44D and SA-44T. From survey year 2016 through survey year 2019, there were eight electronic form types (SA-44, SA-44A, SA-44C, SA-44D, SA-44E, SA-44N SA-44S and SA-44T). Starting with survey year 2020 (which will be collected in 2021), there will only be the two electronic form types named above. Forms SA-44, SA-44A, SA-44C, SA-44E, SA-44N and SA-44S, are being combined with the remaining forms to reduce respondent burden by streamlining data collection operations on the number of forms received by a company. The two remaining worksheets will collect data from companies with and without merchandise lines, enable us to collect information on a NAICS basis, and to request similar data items. Variations in the electronic worksheets are needed to address the size of the firm, kind-of-business, or data items requested.


Each year, estimates generated from the ARTS are released to the public approximately 14 months after the reference period has concluded. These national-level estimates are released (for the various items collected) by NAICS code. The current sample was selected on a 2012 NAICS basis, so the estimates are also released on a 2012 NAICS basis. Data will not be published on a 2017 NAICS basis until the next sample revision occurs, which will not take place during this three-year clearance window. The data are disseminated through the ARTS website. In the future, however, the data will be released via the Census Bureau’s dissemination platform, data.census.gov. The survey year 2020 data products are scheduled to be released through data.census.gov.


The Census Bureau conducts this mandatory survey under the authority of an Act of Congress, Title 13, United States Code, Sections 131, 182, 224, and 225 (see Attachment 5).


2. Needs and Uses


The ARTS serves as a benchmark for the estimates produced from the Census Bureau’s Monthly Retail Trade Survey (MRTS) [OMB No. 0607-0717]. Each year, the MRTS benchmarks to the ARTS to reflect corrections to data for the MRTS and ARTS samples. Corrections are made to replace previously reported data with more accurate data received at a later date or to replace imputed data with reported data obtained from the company. Similarly, the ARTS benchmarks to the results of the Economic Census every 5 years (more can be found on the ARTS benchmarking in the Benchmarking section of OMB Supporting Statement B.) The ARTS respondents have more time to complete their annual surveys than they do for the monthly survey. The ARTS responses are requested at a time when many firms have already compiled audited book figures for their own use. The timing of annual surveys is also such that Census staff can compile more accurate data from independent sources such as a company’s annual report.


The Bureau of Economic Analysis (BEA) uses the data to estimate the change in the private inventories component of gross domestic product (GDP) and output in both the benchmark and annual input-output (I-O) accounts and GDP by industry. Data on sales taxes are also used to prepare estimates of GDP by industry and to derive industry output for the I-O accounts. Data on detailed operating expenses are collected on this survey quinquennially and used to produce national estimates of value added, gross output, and intermediate inputs, and serve as a benchmark for the annual industry accounts which provide the control totals for the GDP-by-state accounts.


The Bureau of Labor Statistics uses the data as input to its Producer Price Indexes and in developing productivity measurements. Private businesses use the estimates in computing business activity indexes.


Other government agencies and businesses use the data to satisfy a variety of public and business needs such as economic market analysis, company performance, and forecasting future demands.


Information quality is an integral part of the pre-dissemination review of the information disseminated by the Census Bureau (fully described in the Census Bureau's Information Quality Guidelines). Information quality is also integral to the information collections conducted by the Census Bureau and is incorporated into the clearance process required by the Paperwork Reduction Act.



3. Use of Information Technology


The ARTS respondents are required to submit their data through the Census Bureau’s online reporting system, Centurion. This system is designed to make the reporting process more flexible, secure, and convenient for respondents. The response-driven nature of the instrument also reduces respondent burden. Respondents can complete the entire survey in one session or work on the survey over several sessions. Centurion automatically saves the entered data when the user proceeds to the next screen, so the users can return later and pick up exactly where they previously left off. The data provided electronically by respondents are immediately available to the Census Bureau.

The initial letter to respondents identifies the survey name, due date, instructions for accessing the survey, authority for collection, and burden estimate. It also provides a telephone number for those needing assistance. The initial and follow-up mailings instruct respondents to access the Census Bureau’s Respondent Portal (https://portal.census.gov). The Census Bureau will also periodically send an e-mail reminder, or eBlast, to respondents. These e-mail reminders provide the same instructions. A telephone follow-up is scheduled for nonrespondents, too. Once the user arrives at https://portal.census.gov/portal, the individual will create a new account (i.e., register) or sign in to an existing account. The Respondent Portal allows users to add and view all current Census Bureau surveys assigned to them. Using the Respondent Portal, respondents can access Centurion (by clicking “Report Now” for the appropriate survey), view their filing status, request extensions, and securely communicate with the Census Bureau.

Once the user logs in to Centurion via the Respondent Portal, the respondent can view PDF worksheets listing all of the questions he or she will be asked to answer. This enables the individual to compile answers to the questions before beginning the electronic reporting process (if desired). Respondents are asked to make note of any special instructions listed either on the main menu or on their worksheet.

The nightly loading of electronic responses allows for a more timely identification of completed questionnaires, which reduces the number of unnecessary follow-up reminders respondents receive.

For the ARTS, electronic reporting is the only advertised reporting option. Nevertheless, the Census Bureau will accept data submitted through other methods. For example, if a company does not have access to the Internet, the Census Bureau can arrange for the business to provide its data to an analyst via telephone.

The table below lists the percentages of submitted questionnaires that were completed via the Internet (Centurion) for the current sample.



20191

2018

2017

2016

98.6%

96.9%

97.3%

96.8%

1As of 9/25/20


4. Efforts to Identify Duplication


Research with other governmental agencies, trade associations, and data users (both government and private sector) via telephone conversations, meetings, trade journal articles, and written correspondence indicates these data are not available from other sources on an ongoing basis.


Monthly sales and inventories data are collected in the Monthly Retail Trade Survey (OMB No. 0607-0717). The monthly survey is voluntary and requests sales, e-commerce, and end-of-month inventories data from a smaller sample of establishments. The mandatory annual program collects more detailed items, and is intended to serve as a benchmark for the monthly sales, e-commerce, and end-of-month inventories estimates. The annual program is also intended to provide estimates on retail trade in the intervening years between the Economic Census years.

5. Minimizing Burden


The Census Bureau reduces the burden placed on respondents by providing several modes of response. While respondents are requested to return their data via the Internet any data prepared and returned on a company's own form is accepted. Respondents are also allowed to provide data directly over the phone. Instructions on the electronic worksheets state that although book figures are preferred, estimates are acceptable.

The Census Bureau attempts to minimize burden by using a stratified random sample design, which uses the least number of sampling units required to produce national level estimates with the desired level of reliability. Data are collected from the largest firms and from a sample of small- and medium-sized firms. The effective sampling rate for the small- and medium-sized firms is 2.3%.

In addition, the Census Bureau attempts to reduce respondent burden by excluding small single-establishment businesses, below a given receipts cutoff, when the questions on the electronic worksheet align with administrative records data. Single unit establishments below a given receipts cutoff, may not be mailed and instead receipts and operating expense data are imputed based on administrative records data.


The Census Bureau makes every effort to minimize the burden on all respondents. The agency evaluates the necessity of each item collected via the survey by interacting with its stakeholders. Centurion, the electronic reporting instrument, is response-driven, which reduces burden by only showing applicable questions to the respondent. The usage of an electronic reporting system with a response-driven design demonstrates the Census Bureau's continued commitment to making the data collection process as easy and reliable as possible. Each year, the survey staff evaluates the design of the reporting instrument and looks for general improvements that can be made in the data collection process. This ensures the data collection minimizes respondent burden as much as possible but still enables the agency to capture accurate information. The Census Bureau also accepts estimates from companies when book figures are not readily available.


Revising the sample approximately every five to seven years also reduces burden. As discussed in Part B (Universe and Respondent Selection section), noncertainty firms from the previous sample are placed in a nonselection category for the new sample if their exclusion would not bias the sample. Overall, the effective sampling rate for the small- and medium-sized firms is approximately 2% of the universe. The Census Bureau staff continuously evaluates the methodological procedures employed for the survey to make sure they are producing reliable and useful information in a way that minimizes burden on respondents. Part B provides additional information regarding the methodology of ARTS, including sampling and benchmarking processes.


6. Consequences of Less Frequent Data Collection


Less frequent data collection of the annual data would result in less accurate monthly sales, quarterly e-commerce and end-of-month inventories estimates since the monthly and quarterly estimates are benchmarked to the annual estimates. Additionally, the GDP would be less accurate due to the loss of annual estimates from the ARTS.


7. Special Circumstances


There are no special circumstances.


8. Consultations Outside the Agency


The Census Bureau issued a pre-submission notice that was published in the Federal Register on Monday, July 6, 2020 located in Vol. 85, No. 129, on pages 40199 and 40200.


We received no comments in response to the pre-submission notice. We did, however, receive a letter of support (Attachment 1) from the Bureau of Economic Analysis. We thank the BEA for their continued support for this collection. The BEA is the Census Bureau’s main source of consultation on the ARTS and regular status meetings and discussions on content are conducted to address any program concerns.

During 2018, a National Academy of Sciences (NAS) panel sponsored by the Census Bureau issued recommendations for the Bureau’s annual economic surveys. The Census Bureau is using these suggestions to improve its surveys.


Because the ARTS will continue to use the same sample of businesses and request essentially the same information during this three-year clearance window, new research was not conducted with respondents to investigate recordkeeping practices, ease of reporting, etc.


9. Paying Respondents


We do not provide any payments or gifts to respondents.


10. Assurance of Confidentiality


The initial letter (see Attachment 2), welcome screen in the electronic reporting instrument (see Attachment 3), and respondent worksheets (see Attachment 4) convey the specific laws that authorize the collection, require response to the survey, and ensure the confidentiality of those submitting information. Specifically, the following block of text is displayed:


Title 13, United States Code, Sections 131 and 182, authorizes this collection. Sections 224 and 225 require your response. The U.S. Census Bureau is required by Section 9 of the same law to keep your information confidential and can use your responses only to produce statistics. The Census Bureau is not permitted to publicly release your responses in a way that could identify your business, organization, or institution. Per the Federal Cybersecurity Enhancement Act of 2015, your data are protected from cybersecurity risks through screening of the systems that transmit your data.


The ARTS website (i.e., the Information for Respondents and FAQs pages) relays similar information to respondents.


Note: For Attachment 2, ARTS-L1, ODYS-RS-L1, ODYS-RW-L1, and ODYS-RWS-L1 are the initial letters the respondents receive. During 2018, a NAS panel issued recommendations for the Census Bureau’s annual economic surveys. The Census Bureau is using these suggestions to improve its surveys. Because of the recommendations, during the 2019 survey year, the Census Bureau started coordinating data collection efforts between the ARTS, Annual Wholesale Trade Survey (AWTS), and Service Annual Survey (SAS). Specifically, if a company was required to submit the ARTS and at least one of these other surveys, the agency identified a single point of contact for the company and sent the individual a letter mentioning each survey that needed to be submitted. (A select group of companies were excluded from this coordinated collection effort and continued to be contacted separately by the relevant surveys. Some of these companies were removed from the coordinated collection efforts prior to the initial mailing based on previous collection experiences and communications with the companies. Other companies reached out to the Census Bureau and asked to be removed at later points in time.) Depending on the surveys the company was selected for, it received either ODYS-RS-L1, ODYS-RW-L1, or ODYS-RWS-L1 If a company was only required to complete the ARTS, it simply received ARTS-L1. Follow-up attempts were also coordinated in a similar fashion. During the next three years, the ARTS will be following a similar outreach approach. Consequently, there are a variety of follow-up messages a respondent may receive. Attachment 2 provides a sample of some of these. The exact wording for the initial and follow-up messages has not been finalized yet, but the final versions should be very similar to the samples. Similarly, the electronic reporting instrument for survey year 2020 is still in the development phase, but the information presented in Attachment 4 should be nearly identical to the information displayed in the final instrument.


All data collected on the ARTS are maintained in the strictest confidence under the authority of an Act of Congress, Title 13, United States Code, Sections 131 and 182, that authorizes this collection. Sections 224 and 225 require a response to this survey. The U.S. Census Bureau is required by Section 9 of the same law to keep all respondent information confidential and can only use survey responses to produce statistics. The Census Bureau is not permitted to publicly release any respondent information in a way that could identify any business, organization, or institution. Per the Federal Cybersecurity Enhancement Act of 2015, all company data are protected from cybersecurity risks through screening of the systems that transmit data to the Census Bureau. (refer to respondent correspondence found in Attachment 2). Additionally, all data reported by companies are immune from legal process. This assurance of confidentiality is conveyed to the company via the respondent worksheets (Attachment 4) and letters (Attachment 2), as well as a statement placed on the login screen for the ECORR reporting portal (Attachment 3). Additionally, the welcome screen of the ECORR provides a statement of authorization at the bottom of the screen (Attachment 3) The letters also inform the respondent that this survey is required by law.



11. Justification for Sensitive Questions


The ARTS does not contain questions of a sensitive nature.


12. Estimate of Hour Burden


The estimated respondent burden is significantly higher in survey year 2022 than in 2020 and 2021 because the ARTS will collect detailed operating expenses in survey year 2022. Detailed operating expenses are collected only in years ending in “2” and “7”. It should also be noted that the estimated respondent burden hours for this annual collection will change over time as sample maintenance activities are performed (refer to the Universe and Respondent Selection section in Part B). Namely, the quarterly birth/death process, removal of out-of-scope cases, and restructuring of reporting units to account for organizational changes (e.g., consolidated collection companies, mergers, acquisitions, etc.) will inevitably cause the number of respondents to fluctuate over time. Historically, the number of respondents typically decreases each year within a given sample. This trend has been true for this sample as well (survey years 2016, 2017, 2018, and 2019). To simplify things for this submission, the estimates provided in this section will assume that the number of respondents remain constant for survey years 2020, 2021, and 2022.

There are an estimated 17,297 number of respondents. The estimated burden is approximately 25,035 hours, computed based on a three-year average of approximately 92 minutes per response. Survey years 2020 and 2021 will be significantly less because the detailed operating expenditures questions will not be a part of the survey during those years. The average time per response is calculated by dividing the average total burden hours over the three-year collection by the total number of respondents per electronic worksheet (see table below). The estimated total annual cost to respondents is approximately $861,204 based on the average response burden of 25,035 hours at approximately $34.40 per hour (according to the median hourly salary for accountants and auditors; Occupational Employment Statistics - Bureau of Labor Statistics May 2019 National Occupational Employment and Wage Estimates; $34.40 represents the national median hourly wage of the full-time wage and salary of accountants and auditors SOC code 13-2011). https://www.bls.gov/ooh/business-and-financial/accountants-and-auditors.htm#tab-1

Survey Years 2020 and 2021


Form

Number of Respondents

Hours

Burden (in hours, Number of Respondents X Hours)

Cost (in dollars, Burden X $34.40)

SA-44D

15,718

0.53

8,331

$ 286,586

SA-44T

1,579

0.72

1,137

$39,113

TOTAL

17,297

NA

9,468

$ 325,699


Survey Year 2022


Form

Number of Respondents

Hours

Burden (in hours, Number of Respondents X Hours)

Cost (in dollars, Burden X $34.40)

SA-44D

15,718

3.23

50,769

$ 1,746,454

SA-44T

1,579

3.42

5,400

$185,760

TOTAL

17,297

NA

56,169

$ 1,932,214


NA = Not applicable



Due to refinements of the estimates for the final submission, there is a slight difference in the burden hour and cost estimates because not all of the data used for the original estimate were available at the time of the preparation of the pre-submission notice.


13. Estimate of Cost Burden

We do not expect respondents to incur any costs other than that of their time to respond. The information requested is of the type and scope normally carried in company records and no special hardware or accounting software or system is necessary to provide answers to this information collection. Therefore, respondents are not expected to incur any capital and start-up costs or system maintenance costs in responding. Furthermore, purchasing of outside accounting or information collection services, if performed by the respondent, is part of usual and customary business practices and not specifically required for this information collection.


14. Cost to Federal Government


The total cost to the Federal Government for the ARTS in fiscal year 2021 is expected to be $2.6 million, all borne by the Census Bureau. The total cost per year is expected to remain the same for the remainder of this authorized collection.


15. Reason for Change in Burden

The burden hour estimate is 5,496 hours less than the previously approved ARTS. This decrease is due to declines in the current sample from 20,067 respondents to 17,297 respondents for this submission. Mergers, acquisitions, other consolidations and market changes due to shifts from traditional brick and mortar shopping to e-commerce has led to significant declines in the number of reporting units in the current ARTS sample.


16. Project Schedule


The initial letter informing regular survey respondents about the survey is sent the second week of March (i.e., the March after the reference year has concluded). For survey year 2020, these letters are scheduled to be mailed on March 11, 2021. The online reporting instrument (Centurion) becomes available to the respondents earlier than the regular survey mailing for those ARTS respondents participating in the coordinated collection combined mailing with the Annual Wholesale Trade Survey (AWTS) and the Service Annual Survey (SAS). (More on the coordinated collection combined reporting and Full Service Account Management respondents are described in the next two paragraphs). Respondents always have at least 30 days to complete the survey online, so the due date typically falls towards the third week of April. Survey year 2020 date will fall on April 20, 2021. Prior to the due date, the agency will send a reminder notification (mailed letter and e-mail) to companies that have not submitted their data yet. If respondents do not submit their data by the due date, the agency continues to remind respondents about the ARTS through various means (e-mails, mailed letters, and/or telephone calls) until they submit the data. Throughout the processing cycle, analysts review the information provided by companies for accuracy. On occasion, they will contact respondents via secure electronic messages (sent through the Census Bureau’s Respondent Portal) or telephone calls to confirm suspicious data. During the fall (beginning of November), the reporting instrument (Centurion) is shut down for the survey year. For survey year 2020, this is currently scheduled to occur on November 1, 2021. Although the electronic collection instrument Centurion is closed, respondents can still provide data to analysts over the telephone. In December, after months of analysis, the company- and industry-level data are finalized for the year. The estimates generated from the ARTS are released to the public approximately 14 months after the reference year has concluded. For survey year 2020, the goal is to release the data by the beginning of February. These national-level estimates are tabulated at the two-, three-, four-, and selected five- and six-digit NAICS levels. Summary data are analyzed to ensure the estimates are consistent with other economic series and meet release standards. Currently, the data are disseminated through the ARTS website. In the future, however, the data will be released via the Census Bureau’s dissemination platform, data.census.gov. The survey year 2020 data products are scheduled to be released through data.census.gov.


As mentioned in the Assurance of Confidentiality section, during 2018, a NAS panel issued recommendations for the Census Bureau’s annual economic surveys. Because of the recommendations, during the 2019 survey year, the Census Bureau started coordinating data collection efforts between the ARTS, AWTS, and SAS. Specifically, if a company was required to submit the ARTS and at least one of these other surveys, the agency identified a single point of contact for the company and sent the individual a letter mentioning each survey that needed to be submitted. (A select group of companies were excluded from this coordinated collection effort and continued to be contacted separately by the relevant surveys because of existing reporting relationships or were later removed from the mailings at the request of the company). Companies that fall under this category are known as “coordinated collection companies.” This coordinated effort will continue during this three-year clearance window. The coordinated collection companies do not follow the exact same collection schedule as the other ARTS cases (which were previously discussed). The initial letter for these coordinated collection cases will be staggered. ARTS cases that share contacts with SAS and AWTS will tentatively be mailed out the third week of January, 2021. Since respondents still receive at least 30 days to submit their data, the due dates will be adjusted accordingly. As of now, it is expected that the due date for the ARTS coordinated collection companies will be no later than the end of February. However, this is subject to change. (The due date reminder will likely be sent at a different point in time, too.) If coordinated collection respondents do not submit their data by the due date, the agency will also continue to remind them about their reporting obligations through various means (e-mails, mailed letters, and/or telephone calls) until they submit the data. Coordinated collection respondents use the same reporting instrument as all other cases. For example, for a coordinated collection company that is required to complete the ARTS and the AWTS, the company will complete the ARTS using the ARTS Centurion instrument. It will then complete the AWTS via the AWTS Centurion instrument. Therefore, the ARTS reporting instrument will close on the same day for all ARTS respondents.


In addition to these changes, there is an added element of Full Service Account Management for approximately 20 large and complex coordinated collection companies across all three surveys. These companies will receive the same schedule as described above but all response follow-up will refer respondents to work directly with their Full Service Account Manager (FSAM) The FSAM program will provide these companies with a coordinated, integrated and unduplicated approach to data collection while aligning to the program area’s reporting requirements. This program establishes a single point of contact for each company at the Census Bureau to discuss and resolve issues. The goal of the FSAM program is to collect timely and accurate response information, provide outstanding customer service, working with respondents and programs to resolve coherence issues, and create customized respondent reporting arrangements.


17. Request to Not Display Expiration Date


We wish to continue to display the expiration date within the online collection system (Centurion).


18. Exceptions to the Certification


There are no exceptions to the certification statement OMB Form 83-I.


19. NAICS Codes Affected


The following are the 3-digit NAICS codes for the businesses affected by this information collection:


NAICS

KIND OF BUSINESS

441

Motor Vehicle and Parts Dealers

442

Furniture and Home Furnishings Stores

443

Electronics and Appliances Stores

444

Building Material & Garden Equipment Dealers

445

Food and Beverage Stores

446

Health and Personal Care Stores

447

Gasoline Stations

448

Clothing and Clothing Accessories Stores

451

Sporting Goods, Hobby, Book and Music Stores

452

General Merchandise Stores

453

Miscellaneous Store Retailers

454

Nonstore Retailers



File Typeapplication/vnd.openxmlformats-officedocument.wordprocessingml.document
AuthorWindows User
File Modified0000-00-00
File Created2021-01-13

© 2024 OMB.report | Privacy Policy