FR3033s_20101229_omb_part_b

Supporting Statement B

Survey of Finance Companies

(FR 3033; OMB No. 7100-0277)

Attachment

Overview of Statistical Methods

Statistical Methods

The Federal Reserve proposes to conduct, with revision, the Census of Finance Companies and Survey of Finance Companies, continuing the regular five-year survey series (the "quinquennial") (FR 3033; OMB No. 7100-0277). This information collection includes both the Census of Finance Companies (FR 3033p) and the Survey of Finance Companies (FR 3033s). The data collected from this voluntary survey would be used to benchmark the consumer and business finance series collected on the monthly Domestic Finance Company Report of Consolidated Assets and Liabilities (FR 2248; OMB No. 7100-0005).

The first stage of the quinquennial, the FR 3033p was implemented in April 2010. The FR 3033p was sent to all companies that met the criteria developed to identify the potential universe of domestic finance companies. An accurate census was required to form a representative sample of finance companies, to which a more detailed survey would be sent. The census gathered limited information including total assets, areas of specialization, and information on the corporate structure of the companies. The Federal Reserve identified approximately 30,000 firms to which the census was sent.

From the universe of finance companies as determined by the FR 3033p, a stratified random sample of 3,000 finance companies would be drawn for the FR 3033s. The survey would be sent in February 2011 and collect detailed information, as of December 31, 2010, from both sides of the respondents' balance sheets, as well as companies’ geographic scope of operations and servicing activities. The data collected from this voluntary survey would be used for two purposes: to benchmark the consumer and business finance series collected on the monthly Domestic Finance Company Report of Consolidated Assets and Liabilities (FR 2248; OMB No. 7100-0005) and to increase the Federal Reserve’s understanding of an important part of the financial system.

The survey would collect balance sheet data on major categories of consumer and business credit receivables and major liabilities. The respondent panel of 3,000 companies would include 45 companies that are FR 2248 reporters and 2,955 non-FR 2248 companies. Respondents would be directed to either complete the survey online or mail responses to the Federal Reserve Board. Once the data from FR 3033s have been analyzed, they would be used as a benchmark to the FR 2248 for the period December 2005 through December 2010.

Given the important changes in the market, the uncertainty about the composition of the previously unknown part of the universe of finance companies, the uncertainty about response patterns within the size groups and specialization categories, and the length of time between benchmark surveys, the Federal Reserve proposes increasing the authorized FR 3033s survey sample size to 3,000 finance companies. Therefore, if approved, the respondent panel size will increase from 750 finance companies to 3,000. As a result of the FR 3033p responses, the Federal Reserve has identified the universe of finance companies for the survey.¹ The survey would be stratified by size of total assets and by specialization in receivables. The survey sample would likely include all FR 3033p respondents that reported at least $1 billion in total assets. For coverage of smaller respondents², the main survey sample would include a sufficient number of companies with assets under $1 billion to provide adequate representation for the expected analytical purposes.

The 2005 population size, total sample, and the number of responses in each stratum are shown in Table 1. Table 1 also shows the estimated response rate associated with two panel sizes (750 and 3,000 companies), based on the response rates to the 2005 survey.³

Given that the universe of finance companies has probably changed significantly since December 2005, error estimates based on the 2005 data must be interpreted as a rough guideline for the census panel structure. The Federal Reserve has considered the error implications associated with two panel sizes (750 and 3,000 companies) and their associated response patterns.

To maximize the response rate, the timing of the FR 3033s would coincide as closely as possible to fourth-quarter SEC 10-K filings. The Federal Reserve believes that by targeting the time period just after the 10-K filings, it is more likely that respondents will have the necessary information readily on hand to complete the FR 3033s. The survey would be mailed after February 15, 2011, with instructions to respond by March 15, 2011. After March 15, 2011, follow-up mailings would be sent to companies that do not respond and follow-up calls would be made after April 1, 2011.

Table 1

Survey Response Distribution by Asset Size of Finance Company

Asset size	2005 FR 3033p	2005 FR 3033s		2010 Estimated response for FR 3033s panel size
	Estimated population	Total sam­ple	Number of usable re­sponses	750	3000
Less than $1 million	381	18	3	33	182
$1 to less than $10 million	574	332	41	24	104
$10 to less than $100 million	347	219	52	36	179
$100 million to less than $1 billion	131	98	25	32	64
$1 to less than $3 billion	23	23	10	10	10
$3 to less than $20 billion	34	31	11	11	11
$20 billion or more	19	19	19	19	19
Totals	1,509	740	161	165	569

The estimated response distributions, shown in Table 2, assume that the response rate for each size and specialization combination would be no greater than the level attained in 2005. The Federal Reserve calculated the standard errors of the simple expansion estimator that is used to derive population aggregates under the assumption that whether a company responded or not was completely random. Strictly speaking, the standard errors for sample sizes (assuming the response rate mentioned above) can only be said to hold for the 2005 survey population. Response rates may fall below the level of 2005, but the Federal Reserve intends to make a concerted effort to raise response rates substantially; the estimated standard errors would fall if this effort is successful.

Table 2

Estimated Standard Errors of Aggregate Levels of Finance Company

Receivables in Selected Balance Sheet Categories for Alternate Panel Sizes

(Based on 2005 benchmark data; $ in billions with % of outstanding in parentheses)

Financial items	Panel size
	750	3,000
	Estimated number of responses
	165	569
Total	46.67 (2.12)	46.01 (2.09)
Total consumer	23.82 (3.88)	23.32 (3.80)
Motor vehicle	37.51 (11.10)	37.49 (11.10)
Revolving	3.07 (4.57)	3.05 (4.54)
Other	16.67 (8.91)	15.84 (8.47)
Real estate	16.60 (3.12)	16.32 (3.07)
Total business	29.68 (5.86)	29.51 (5.83)
Motor vehicle	7.93 (8.09)	7.12 (7.26)
Equipment	32.32 (12.75)	32.12 (12.67)
Other	3.80 (4.00)	3.21 (3.38)
Bank loans	7.81 (8.89)	6.77 (7.71)

Note: For calculating standard errors, some combinations of size group and specialization category have been collapsed into a single size group when the sample size was insufficient.

1 The Federal Reserve has the following pre-determined set of characteristics to internally determine if the respondent is a finance company. A finance company is defined as (1) a domestic entity, (2) a company with 50 percent or more of its assets in the form of loans or leases, and (3) a nondepository institution.

2 For the purposes of this proposal, the term small or smaller respondent or company is defined as a company that reports less than $1 billion in assets.

3 The Federal Reserve applied the comparable 2005 response rates for each size and specialty group so that the overall response rate estimated for the 2010 survey need not equal the overall response rate of 22 percent in 2005.