Home > Deposit Insurance > The Deposit Insurance Funds > Household Survey on Deposit Insurance Awareness, April 2001




Household Survey on Deposit Insurance Awareness, April 2001


Survey Report

Prepared by:
The Gallup Organization
Government Division
901 F Street, N.W.
Suite 400
Washington, D.C. 20004
202.715.3030

Authors:
Darby Miller Steiger
Alison Simon
Robert Montgomery
TABLE OF CONTENTS

Introduction
Executive Summary
Detailed Findings
1. Awareness of FDIC and FDIC-Insured Products & Services
2. Preference for Obtaining Information About FDIC Deposit Insurance
3. Preference for Obtaining General Deposit Insurance Information
4. Investing Decisions and FDIC Insurance
5. Support for Changing the Insured Deposit Level
6. Current Banking Habits
Technical Notes
Sample Selection
Nonresponse
Weighting
Imputation

 

Introduction

The Federal Deposit Insurance Corporation (FDIC), as the agency statutorily responsible for insuring deposits at banks and savings institutions, has conducted a comprehensive review of the deposit insurance system. The intent was to ensure that the system continues to meet the publicís need for a safe place to invest its savings, and that federal deposit insurance will continue to contribute fully to the stability of the U.S. banking system.

To assist in its analysis of deposit insurance reform issues, the FDIC contracted with The Gallup Organization to conduct a national household survey. The purpose of the survey was twofold: (1) To collect information regarding the publicís understanding of federal deposit insurance and depositorsí need for information about their insurance coverage; and (2) to gather information about the publicís deposit insurance coverage needs and the possible effects of increasing the deposit insurance level.

The FDIC planned to use data from the survey to assist in its ongoing efforts to educate the public about federal deposit insurance coverage. It also planned to use the survey data to develop recommendations regarding the manner by which coverage levels are established. These efforts were intended to ensure that depositors have the information they need to take full advantage of their insurance protection and recognize what is, and is not, insured at financial institutions.

The Gallup Organization conducted telephone interviews with a randomly selected, representative sample of 1,658 adults who identified themselves as the people most knowledgeable about household finances (household Chief Financial Officers, or CFOs), age 18 or older, living in households with telephone service in the continental United States. The interview period ran from November 20 to December 23, 2000.

After interviewing was completed, the data were weighted to adjust for nonresponse. A detailed description of the weighting methods can be found in the Technical Notes.

All sample surveys are subject to various sources of error, generally classified as sampling error and nonsampling error. Sampling error relates to deviations that arise because a sample is used to generalize to the universe. Sources of nonsampling errors include undercoverage of the target population, survey nonresponse, and measurement errors. Additional discussion of survey errors is given in the initial portion of the Technical Notes.

In the remainder of this report, we describe in detail the weighted responses to the national survey of public attitudes and knowledge about FDIC deposit insurance.

Executive Summary

The Gallup Organization found that, in general, the vast majority of household Chief Financial Officers, or CFOs (those most knowledgeable about the householdís finances), knew of the FDIC and whether their banks were insured. However, they lacked specific knowledge about whether certain transactions were insured and the details of joint account coverage. Most preferred to obtain information about deposit insurance directly from their financial institution, although those looking to the FDIC for information favored obtaining information from the FDIC's Web site. The survey also found that FDIC coverage is an important factor in investment decisions. Yet, while most CFOs indicated they would like to see the deposit insurance level keep pace with inflation, they were split as to whether the level should be raised -- about one-fourth believing the current level is appropriate, another one-fourth having no opinion, and about half believing the level should be raised.

1. Awareness of FDIC and FDIC-Insured Products & Services

FINDING: In general, household CFOs were well acquainted with the name "FDIC" and knew whether their financial institutions were insured by the FDIC. The CFOs, however, did not know the specifics of FDIC insurance coverage of various banking products and services.

  • Overall, more than eight in 10 household CFOs have heard of the FDIC (an estimated 85 percent) and another eight in 10 knew whether their banks were insured or not (83 percent).
  • The public is much less knowledgeable about the specific financial transactions that are insured by the FDIC than they are about whether their own financial institution is insured. When asked the basic amount of money the FDIC insures, just half of the public surveyed correctly identified $100,000 as the basic level (49 percent). About one-third readily admitted that they did not know the answer to that question (35 percent).
  • Awareness of specific transactions insured by the FDIC is also low. Only a slight majority was aware that the FDIC does not insure all bank transactions (57 percent); the remainder either believed that all transactions are covered (27 percent) or did not know enough to say (16 percent).
  • Furthermore, many do not know that certain types of investments are not covered by FDIC insurance, including insurance annuities (63 percent either donít know or believe they are insured), mutual funds (56 percent), stocks & bonds (50 percent), and/or Treasury bills (75 percent).
  • When asked whether they knew specifics on joint account coverage, the lack of knowledge was even greater. About three-fourths (75 percent) incorrectly believe that only $100,000 of a joint account with $200,000 in the account is covered, and only 16 percent correctly know that the full amount is covered. Furthermore, three-fourths do not realize that individual accounts are added together and insured to $100,000 (56 percent believe accounts are insured separately to $100,000 and another 20 percent do not know).

2. Preference for Obtaining Information About FDIC Deposit Insurance

FINDING: The public would rather get information from banks than directly from the FDIC.

  • When asked where they prefer to get their information about deposit insurance, roughly six in 10 reported a preference for receiving information from their bank (57 percent). Another three in 10 reported a preference for receiving information from the FDIC (33 percent).
  • Those who prefer to get information from their bank are most apt to talk directly to a bank representative (58 percent would be very likely to do this), followed closely by reading information included in their account statement (55 percent very likely). Slightly fewer indicated they would be very likely to read a brochure, advertisement or Web site from their bank (43 percent very likely), and only 8 percent would be likely to watch a video in their bankís lobby.
  • Among those who prefer to get information about deposit insurance directly from the FDIC, the top preference was to visit the FDICís Web site (see Figure 7). Four in 10 of those who would go to the FDIC for information said they would be very likely to visit the FDICís Web site (42 percent). The public was less likely to call the FDIC (28 percent very likely) or write the FDIC (14 percent).

3. Preference for Obtaining General Deposit Insurance Information

FINDING: In spite of a lack of knowledge about details of deposit insurance, many feel they have enough information.

  • Over one-half of household CFOs believe they have a sufficient amount of information on deposit insurance (an estimated 55 percent), while 42 percent would like more information.

4. Investing Decisions and FDIC Insurance

FINDING: FDIC coverage is a significant factor in investment decisions, especially when household

wealth is taken into account.

  • Among household CFOs, about one-third said that at some point they have inquired about whether the FDIC insured an investment before making an investment decision (34 percent), while 66 percent indicate they have never inquired.
  • Among those who said they have inquired about whether an investment was insured, the inquiry most frequently was directed to the financial institution in which they were investing (64 percent). A smaller group has also asked their investment advisor about this (28 percent), and just 2 percent have gone directly to the FDIC.
  • Among those who have invested money in the past five years, additional questions were posed to determine the relative importance of FDIC insurance in an investment decision. To recent investors, the two most important factors are the potential return on investment (75 percent rated this as very important) followed closely by the risk of the investment (71 percent). While fewer investors were concerned that their investment was fully protected by loss by the FDIC (57 percent), FDIC coverage is still very important to over half of recent investors.

5. Support for Changing the Insured Deposit Level

FINDING: The public would like to see the deposit insurance level keep pace with inflation, but is split over whether the level should be raised or not.

  • Households were asked if they preferred to leave the deposit insurance level alone, or whether they felt it should be adjusted over time to keep pace with inflation. Support is much stronger for adjusting the level over time, with over three-quarters agreeing that the level should keep pace with inflation (77 percent). Just 16 percent believed the level should be left alone.
  • When asked if they believed the level of FDIC deposit insurance is at the right level now or if it needs to be raised, only 27 percent believed that the insurance level is at the appropriate level today. Nearly half believed the level should be raised (47 percent), with an additional 25 percent having no opinion.
  • More specifically, 27 percent of the public feels the level is at the right level now, 5 percent feel it should be increased to $150,000, 21 percent said it should be increased to $200,000, and an additional 19 percent said it should be raised to above $200,000. The remainder were uncertain.
  • Regardless of any economic or personal scenario, if the deposit insurance level were raised, the public is somewhat willing to shift more money into insured deposits. Only 23 percent would be very likely to shift more money, with another 36 percent feeling "somewhat" likely.

6. Current Banking Habits

FINDING: Although very few admit to keeping over $100,000 in the bank, household CFOs did admit to keeping funds at different banks to make sure their money is fully insured.

  • Among CFOs who have recently put money aside for savings or investing, a plurality keep less than one-quarter of their funds in banks (an estimated 46 percent). An additional 27 percent keep between one-fourth and one-half of their funds in a bank.
  • Just 12 percent reported keeping more than $100,000 in the bank, with most of those keeping between $100,000 and $200,000 in the bank (66 percent).
  • While few currently have large amounts of money in the bank, about one-third of CFOs reported having more than $100,000 in the bank at one time or another (30 percent). Reasons for having large deposits included either selling a home (64 percent), receiving an inheritance (25 percent), or some other large financial transaction (31 percent).
  • Large depositors who said they did currently have over $100,000 in a bank (12 percent of the population) were asked if any of that money was currently uninsured. One third of these large depositors said they had uninsured deposits (33 percent), while most others were under the impression that all of their money was insured (63 percent). The amount they believe is uninsured varied from less than $50,000 (38 percent) to between $50,000 and $100,000 (23 percent) to $100,000 or more (39 percent).
  • Among large depositors, 58 percent said they keep money at more than one bank to make sure that their money is fully insured, whereas 42 percent do not. Among those large depositors who do keep their money at more than one bank to stay fully insured (7 percent of the population), more than six in 10 (63 percent) also report that they keep their deposits at any one bank to just under $100,000 in order to stay fully insured.

Detailed Findings

1. Awareness of FDIC and FDIC-Insured Products & Services

Overall, more than eight in 10 household CFOs have heard of the FDIC (an estimated 85 percent) and another eight in 10 knew whether their banks were insured or not (83 percent). Hispanics were among the least likely to have heard of the FDIC (59 percent), as were African Americans (68 percent) and other minorities (66 percent). Young adults aged 18 to 25 (68 percent), as well as those with less than a high school diploma (61 percent) were also less likely to have heard of the FDIC.

The public is much less knowledgeable about the specific financial transactions that are insured by the FDIC than they are about whether their own financial institution is insured. When asked the basic amount of money the FDIC insures, just half of the public surveyed correctly identified $100,000 as the basic level (an estimated 49 percent). About one-third readily admitted that they did not know the answer to that question (35 percent) (see Figure 1). Senior citizens and the wealthy (those with income over $75,000) were most likely to know the basic FDIC insurance level (59 percent and 60 percent, respectively), while young adults under age 25 were among the least aware (61 percent donít know). Large depositors, who indicated they had more than $100,000 in the bank, were much more likely to know the basic level (67 percent) than were those who had less than $100,000 in the bank (46 percent).

Figure 1 - Bar chart conveys graphically the information presented in the previous paragraph.   What is the basic amount of money the FDIC will insure?

Awareness of specific transactions insured by the FDIC is also low. Only a slight majority was aware that the FDIC does not insure all bank transactions (57 percent); the remainder either believed that all bank transactions are covered (27 percent) or did not know enough to say (16 percent). Large depositors were much more likely to know that not all transactions are insured (70 percent) than were smaller depositors with less than $100,000 in the bank (55 percent). Interesting regional differences emerged on this question. Northeasterners (67 percent) and Westerners (62 percent) were much more likely to know that only some transactions are insured than were their counterparts in the Midwest (52 percent) and South (51 percent) (see Figure 2).

Figure 2 - Bar chart conveys graphically the information presented in the previous paragraph.  Do you think all transactions at banks are FDIC insured, or are only some of them insured?

Furthermore, many do not know that certain types of investments are not covered by FDIC insurance, including insurance annuities (63 percent either donít know or believe they are insured), mutual funds (56 percent), stocks & bonds (50 percent), and/or Treasury bills (75 percent) (see Figure 3).

Figure 3 -Bar chart conveys graphically the information presented in the previous paragraph.  Are any of the following types of investments insured by the FDIC?

When asked whether they knew specifics on joint account coverage, the lack of knowledge was even greater. Three-fourths (an estimated 75 percent) incorrectly believe that only $100,000 of a joint account with $200,000 in the account is covered, and only 16 percent correctly know that the full amount is covered (see Figure 4). Furthermore, three-fourths do not realize that individual accounts are added together and insured to $100,000 (56 percent believe accounts are insured separately to $100,000 and another 20 percent do not know). The lack of knowledge of specific rules on FDIC insurance levels applies to old and young, rich and poor, male and female, large depositors and small depositors.

Figure 4 -Bar chart conveys graphically the information presented in the previous paragraph.  Knowledge of insurance rules regarding separate insurance of different account types.

There appears to be a need for further education of the public. In addition to uncertainty about FDIC insurance levels, consumers are also unsure about the more specific rules that distinguish between insured deposits and other bank products, as well as the rules regarding the separate insurance of different account types. Rather than acknowledging they do not know the rules, an overwhelming majority of consumers gave wrong answers to fundamental questions about which products are insured by FDIC and about coverage levels.

2. Preference for Obtaining Information About FDIC Deposit Insurance

When asked where they prefer to get their information about deposit insurance, roughly six in 10 reported a preference for receiving information from their bank (57 percent). Another three in 10 reported a preference for receiving information from the FDIC (33 percent) (see Figure 5). No subgroups stood out with stronger preferences, though Midwesterners were slightly more likely to want information from their bank (63 percent), while Northeasterners and non-African American minorities were more likely to prefer getting information directly from the FDIC (38 percent and 39 percent, respectively).

Figure 5 - Bar chart conveys graphically the information presented in the previous paragraph. Where would you prefer to get information about deposit insurance

Those who prefer to get information from their bank are most apt to talk directly to a bank representative (58 percent would be very likely to do this), followed closely by reading information included in their account statement (55 percent very likely). Slightly fewer indicated they would be very likely to read a brochure, advertisement or Web site from their bank (43 percent very likely), and only 8 percent would be likely to watch a video in their bankís lobby (see Figure 6). Among those who the FDIC might judge as most in need of information, that is, those who do not know what the basic insurance level is, the top preference is also to talk to a banker (57 percent), although this group shows less interest in reading information in their account statement (42 percent, compared to 55 percent overall).

Figure 6 - Bar chart conveys graphically the information presented in the previous paragraph.   If you wanted to get information about deposit insurance from your bank, how likely would you be to do each of the following?

Among those who prefer to get information about deposit insurance directly from the FDIC, the top preference was to visit the FDICís Web site (see Figure 7). Four in 10 of those who would go to the FDIC for information said they would be very likely to visit the FDICís Web site (42 percent). The public was less likely to call the FDIC (28 percent very likely) or write the FDIC (14 percent).

Figure 7 - Bar chart conveys graphically the information presented in the previous paragraph.  If you wanted to get general information about deposit insurance from the FDIC, how likely would you be to do each of the following?

3. Preference for Obtaining General Deposit Insurance Information

FINDING: In spite of a lack of knowledge about details of deposit insurance, many feel they have enough information.

Over one-half of household CFOs believe they have a sufficient amount of information on deposit insurance (an estimated 55 percent), while 42 percent would like more information (see Figure 8). Those who were most likely to want additional information include African Americans (64 percent), Hispanics (52 percent), and young adults aged 18-25 (49 percent); senior citizens are most likely to believe they have enough information (70 percent). There is no clear relationship between income level and preference for obtaining additional information. However, those with large amounts of savings and investments (over $250,000) are less apt to want information (33 percent) than those with savings and investments under $250,000 (44 percent)

Those who know that the basic insurance level is $100,000 are more apt to say they have enough information (63 percent) than those who do not know the insurance level (55 percent).

Figure 8 - Bar chart conveys graphically the information presented in the previous paragraph.  Do you feel you have enough information about FDIC deposit insurance, or would you like more information?

4. Investing Decisions and FDIC Insurance

FINDING: FDIC coverage is a significant factor in investment decisions, especially when household

wealth is taken into account.

Among household CFOs, about one-third said that at some point they have inquired about whether the FDIC insured an investment before making an investment decision (34 percent), while 66 percent indicate they have never inquired (see Figure 9). Senior citizens (40 percent) and those aged 55 to 64 (40 percent) are the most likely to have made this inquiry, as are those with savings or investments of over $500,000 (44 percent) and those with large bank deposits (55 percent).

Figure 9 - Bar chart conveys graphically the information presented in the previous paragraph.   Have you ever inquired about whether an investment was insured by the FDIC before you made the investment?

Among those who said they have inquired about whether an investment was insured, the inquiry most frequently was directed to the financial institution in which they were investing (64 percent). A smaller group has also asked their investment advisor about this (28 percent), and just 2 percent have gone directly to the FDIC.

Among those who have invested money in the past five years, additional questions were posed to determine the relative importance of FDIC insurance in an investment decision. To recent investors, the two most important factors are the potential return on investment (75 percent rated this as very important) followed closely by the risk of the investment (71 percent). While fewer investors were concerned that their investment was fully protected by loss by the FDIC (57 percent), FDIC coverage is still very important to over half of recent investors (see Figure 10).

Figure 10 - Bar chart conveys graphically the information presented in the previous paragraph.   How important is each factor to you when deciding where to invest or put your household's money?

Interesting differences emerged across the social and economic segments, primarily among responses to the importance of the FDIC protecting the investment. Women were less concerned about the investment being protected by the FDIC than were men (49 percent very important among women versus 64 percent among men); the older the household CFO, the more important was FDIC insurance when deciding where to invest; and the lower the income level, the more important was FDIC protection. Similarly, the lower the householdís savings and investment level, the more important FDIC insurance was in the investment decision. In sum, while FDIC insurance is clearly not the most important factor in making investment decisions, it does play a role, particularly for more risk-averse investors such as older and less affluent households.

5. Support for Changing the Insured Deposit Level

FINDING: The public would like to see the deposit insurance level keep pace with inflation, but is split over whether the level should be raised or not.

Households were asked if they preferred to leave the deposit insurance level alone, or whether they felt it should be adjusted over time to keep pace with inflation. Support is much stronger for adjusting the level over time, with over three-quarters agreeing that the level should keep pace with inflation (an estimated 77 percent) (see Figure 11). Just 16 percent believed the level should be left alone. Support for changing the level increases with education level, income level, and savings/investments level.

Figure 11 -Bar chart conveys graphically the information presented in the previous paragraph.   Do you think the limit on FDIC deposit insurance should be adjusted over time to pace with inflation, or should it be left alone?

When asked if they believed the level of FDIC deposit insurance is at the right level now or if it needs to be raised, only 27 percent believed that the insurance level is at the appropriate level today. Nearly half believed the level should be raised (47 percent), with an additional 25 percent having no opinion. Women were more supportive of raising the level than were men (55 percent versus 41 percent).

Support for raising the insurance level is closely correlated with income. Among households earning less than $25,000, 34 percent believe the level needs to be raised; among households earning $75,000 or more, 60 percent feel it needs to be raised (see Figure 12).

Figure 12 - Bar chart conveys graphically the information presented in the previous paragraph.   Do you think the insurance limit is at the right level now, or does it need to be raised?

In addition, as a householdís level of savings and investments rises, so too does support for raising the insurance level. Just 35 percent of those with less than $50,000 in savings and investments believe the insurance level needs to be raised, compared to an estimated 62 percent of those with investments over $250,000. Furthermore, those with large deposits of $100,000 or more in the bank are much more supportive of an increase in the insurance level (67 percent) than are those with smaller amounts in the bank (45 percent). There is also some indication of regional differences, with Northeasterners being more supportive of an increase (54 percent) than either Midwesterners (43 percent) or Southerners (45 percent).

Among household CFOs who indicated the level needs to be raised, 84 percent believe it should be raised to at least $200,000, if not higher. However, since only 47 percent of the sample was asked this question, only about 40 percent of the public believes the amount should be raised to at least $200,000 (see Figure 13). More specifically, 27 percent of the public feels the level is at the right level now, 5 percent feel it should be increased to $150,000, 21 percent said it should be increased to $200,000, and an additional 19 percent said it should be raised to above $200,000. The remainder were uncertain.

Figure 13 - Bar chart conveys graphically the information presented in the previous paragraph.   Is the FDIC insurance limit at the right level now?  If not, what do you think the FDIC Insurance limit sould be raised to?

Another way of understanding support for a potential level change was to present two scenarios and ask the public how they would react. These scenarios may also help gauge the effect of a potential level increase. In this manner, the public is not asked for a viewpoint, but rather how they would respond to a given situation. It should be noted that asking people to estimate a response given a "potential" scenario is quite different than a traditional opinion question. Because people do not necessarily act as they say they will, a personís estimate of how they may respond in a given situation is tenuous at best. Therefore, interpretation of the data may be less reliable for these measures.

In the first scenario, the public was told that either the stock market crashes or there is a recession; would they be prompted to move their money to insured deposits at banks or savings associations or wouldnít it make a difference? Overall, six in 10 said they would be more likely to move their investments (60 percent), while 37 percent said it would not make a difference. When those who said it wouldnít matter were asked if their answer would change if the deposit insurance level increased, only 21 percent said it would (8 percent of the total population). Thus overall, 68 percent of the public said they would move their money to insured deposits if there were an economic downturn or if the insurance level were raised in the midst of an economic downturn. Those who were more likely to change their behavior because of an increase in the deposit insurance level were Northeasterners, those living in urban areas, African Americans and other racial minorities, Hispanics, young adults under 25, and large depositors.

In the second scenario, the public was told that the economy and the stock market are stable, but that they were getting close to retirement and would need to live on their retirement income. In this scenario, would the household CFO be any more likely to move their money to insured deposits? Identical to the first scenario, six in 10 said they would be more likely to move their investments (61 percent). Young adults age 18 to 25 were the most likely to be willing to move their investments (70 percent). Among the indifferent (35 percent of the population), if the deposit insurance level was increased, 27 percent said they would be more likely to move their investments if the deposit insurance level were raised (9 percent of the total population). Thus overall, 70 percent of the public would move their money to insured deposits if they were getting close to retirement, or if the insurance level were raised at a time when they were getting close to retirement.

Regardless of any economic or personal scenario, if the deposit insurance level were raised, the public would be somewhat willing to shift more money into insured deposits. Only 23 percent would be very likely to shift more money, with another 36 percent feeling "somewhat" likely. Few subgroup differences emerged on this measure, but large depositors (who currently have more than $100,000 in the bank) indicated a greater willingness to move their money into insured deposits should the level be raised than did smaller depositors (68 percent very or somewhat likely versus 57 percent for smaller depositors) (see Figure 14).

Figure 14 - Bar chart conveys graphically the information presented in the previous paragraph.   If the limit on FDIC insurance were raised, how likely would you be to put more of your household's money into insured bank deposits?

When asked if they would be more likely to switch funds to a smaller bank with a level increase, only two in 10 said they would (22 percent). Those living in rural areas were more apt to want move their funds (29 percent more likely) than those living in urban (24 percent) or suburban areas (19 percent). Large depositors also indicated a greater willingness to move their deposits to smaller institutions (31 percent).

6. Current Banking Habits

FINDING: Although very few report keeping over $100,000 in the bank, the public does admit to keeping funds at different banks to make sure their money is fully insured.

Among CFOs who have recently put money aside for savings or investing, a plurality keep less than one-quarter of their funds in banks (an estimated 46 percent). An additional 27 percent keep between one-fourth and one-half of their funds in a bank. In general, as a householdís level of savings and investments rises, a household is more likely to keep a smaller proportion of their money in the bank, though the actual amount they have in the bank is higher as wealth increases (see discussion below). Two-thirds of those with more than $500,000 in savings and investments keep less than one-quarter of that money in banks (69 percent), compared to 51 percent of those with $101,000 to $250,000 and 42 percent of those with under $50,000 in investments (see Figure 15).

Figure 15 - Bar chart conveys graphically the information presented in the previous paragraph.   What proportion of your household's savings and investments do you keep in the bank?

Just 12 percent report keeping more than $100,000 in the bank, with most of those keeping between $100,000 and $200,000 in the bank (66 percent). Senior citizens (18 percent) and upper income households (21 percent) are among the most likely to have more than $100,000 in the bank.

While few currently have large amounts of money in the bank, roughly one-third of CFOs admitted to having more than $100,000 in the bank at one time or another (30 percent). Reasons for having large deposits included either selling a home (64 percent), receiving an inheritance (25 percent), or some other large financial transaction (31 percent).

Those who said they did currently have over $100,000 in a bank (12 percent of the population) were asked if any of that money was currently uninsured. One third of these large depositors said they had deposits in the bank that were uninsured (33 percent), while most others were under the impression that all of their money was insured (63 percent). The amount they believe is uninsured varied from less than $50,000 (38 percent) to between $50,000 and $100,000 (23 percent) to $100,000 or more (39 percent).

Among large depositors, 58 percent said they keep money at more than one bank to make sure that their money is fully insured, whereas 42 percent do not. Among those large depositors who do keep their money at more than one bank to stay fully insured (7 percent of the population), more than six in 10 (63 percent) also report that they keep their deposits at any one bank to just under $100,000 in order to stay fully insured.

A question on treasury bills revealed that about four in 10 household CFOs have considered purchasing a U.S. Treasury security (44 percent) and of those, 30 percent said they would be more likely to buy an insured CD if there were fewer securities to buy in the future. Among those for whom it would not make a difference (65 percent, or 29 percent of the total population), 28 percent would be more likely to buy an insured CD if the level of deposit insurance increased. Thus, overall, 38 percent of the public would be willing to buy an insured CD if fewer Treasury securities were available, or if the insurance level increased when fewer Treasury securities were available.

 

Technical Notes

This section of the report describes the methodology used in conducting the survey and producing estimates based on the survey results. It covers the sample selection, administration, nonresponse, weighting, and imputation procedures.

Gallup conducted telephone interviews with a national sample of 1,658 non-institutionalized adults in telephone households in the United States. Data collection was conducted between November 20, 2000 and December 23, 2000, with a four-day pause during the Thanksgiving holiday.

All sample survey estimates are subject to a variety of sources of error. For example, with any randomly selected sample, the sample statistics may deviate from the corresponding population figures because of random sampling error. In addition, the sample results may be distorted by the effects of noncoverage of some portion of the population, by the impact of nonresponse on the survey results, and by errors introduced in the measurement process. In this survey, roughly 8 percent of the household population were omitted by the sample design; most of these omissions involve households without telephones. Another potentially serious source of error in the results is nonresponse. In general, the bias in means and proportions due to nonresponse is a product of the nonresponse rate and the average difference between the respondents and nonrespondents. The CASRO response rate for this survey (30 percent) leaves room for potential nonresponse biases. Below, we provide more detailed information about the sample selection procedure (and its omissions), nonresponse, and the weighting procedure (which attempted to offset the effects of noncoverage and nonresponse).

Sample Selection

The sample was selected in two steps. In the first step, a sampling frame of telephone numbers was defined and stratified based on two income levels. In the second step, a systematic national sample of random telephone numbers was generated for each stratum. Once the sampling was complete, we attempted to contact these numbers and speak with the Chief Financial Officer (CFO) of the household.

The survey instrument was intended to gather FDIC insurance level opinions from households. Currently, these levels are most relevant to households in the higher income ranges (as they would be more likely to keep this much money in an FDIC insured institution). Therefore, in order to ensure sufficient responses to the survey questions, a higher income group was oversampled.

The stratification of the income levels was identified and communicated to Gallupís regular vendor, Survey Sampling, Inc (SSI). SSI maintains a Master Exchange File that contains demographic estimates at the area code and three-digit prefix level. Along with information such as the number of working blocks, counts of listed residential and commercial telephone numbers, the plurality ZIP code, and plurality FIPS code, the file includes demographics such as ethnic densities, urban classification, and mean income. To select a random digit Targeted Income Sample, SSI computes an average of the income predictor scores (scores assigned based on census tract information) at the household level for each telephone exchange. Then the exchanges are ranked by predicted income. With the income defined, SSI then specified a level so that the oversample would be selected only from those exchanges where the average of the income predictor scores were calculated to be at that level or higher.

The average income of $65,000 was chosen jointly by Gallup and FDIC. With the strata defined as such, the proportion of exchanges in the higher income stratum was 14%. Due to the oversampling, the ratio of sampling proportions for the two strata was 5:1. Although the average income of $65,000 was used to define the strata, household incomes exceeding $65,000 would occur in both strata.

The second step was to select a list-assisted telephone sample from each of the two income strata ($65K and more and under $65K). Gallup typically uses SSIís "Random A" methodology to select phone numbers for their RDD samples. The list frame consists of all possible 10-digit telephone numbers in working banks with three or more listed telephone numbers (commonly known as a 3+ design).

A bank is a group of 100 consecutive numbers that share their first eight digitsóthat is, their area code, exchange, and the first two numbers of their four-digit suffixes. For example, all the possible telephone numbers beginning 301-838-57_ _ form a single bank. Prior to selection, the banks are classified according to the number of residential listings they contain. In the method developed by Casady and Lepkowski (1993), numbers from the stratum of banks containing no residential listings ("zero" banks) are under-sampled to reduce the number of calls to unassigned or non-working numbers. Using a "truncated" design, the sample omitted banks with two or fewer residential listings.

A list-assisted, 3+ design would yield total undercoverage of the household population of about 8.4% (6% for non-telephone households and another 2.4% of households with phones that are missed by the frame). The omission of these banks sharply increases the proportion of working residential numbers in the sample (from approximately 25 percent to about 50 percent, depending on the exact rule for dropping banks). Evaluations of the bias associated with the omission of such banks indicate that it is small (Brick, Waksberg, Kulp, & Starer, 1995; Giesbrecht, 1997).

In total, the targeted sample size was 1,920 completes. This number (1,920) was chosen from the design specifications that outlined a sample such that a proportion for any subgroup that makes up 25% of the universe can be estimated to within + 5 percentage points (or less) with a 95% confidence interval.

In total, 12,000 numbers were ordered from SSI to account for unassigned and business numbers (6,615 low income, 5,385 high income in total ordered). Before turning over the sample to Gallup, SSI ran the sample numbers through an auto-dialing procedure that detected (and eliminated) some of the unassigned and nonworking numbers, as well as modem and fax numbers. Approximately one-half of these numbers can be identified by this process, improving the working phone rate of RDD sample by an average of 10-15%. The remaining numbers were then run through SSIís Business Number Purge. In general, an average RDD sample will contain between 12-15% business numbers and this process identifies about one-half of these business numbers. No other numbers were excluded from the sample (for example, numbers that had already been selected by SSI for some prior studies were not excluded).

The sample SSI provided to Gallup included 9,890 numbers. With an assumed working residential number rate of 50 percent, and a target response rate of 50%, it was anticipated that 9,890 numbers would be adequate to provide 1,920 completed interviews (this number accounts for the assumed introduction of a design effect due to the oversampling of about 1.25).

For quality control purposes during interviewing, Gallup creates and releases replicates of 500 telephone numbers. In total, all 9,890 numbers were used during this study (see sample disposition table below).

Gallup then attempted to contact persons in households linked to the sample telephone numbers. When a person answered, telephone interviewers verified that the number was linked to a residence and attempted to select the eligible adult (at least 18 years old). Gallup asked to speak with the adult in the household most knowledgeable about the householdís finances. If no such person existed or if the selected respondent volunteered that the household does not keep any money in a bank or savings association, the household was not eligible to participate in the study. Roughly two percent of the contacted sample was determined to be ineligible.

Nonresponse

While the field period was short and included the Thanksgiving holiday weekend, during which no interviewing was conducted, several efforts were made to maximize response rates on this study. First, a reverse directory look-up was performed on the sample before fielding the study to obtain addresses for as many households as possible. Addresses were found for 35 percent of the original sample frame (n=9,890). Pre-notification letters signed by FDIC Chairman Donna Tanoue were mailed to 4,200 households on November 20, 2000, explaining that a Gallup interviewer would be calling in the next few days, and detailing why the householdís participation was important.

Second, an extensive call design was implemented for the study to increase the probability of contact and cooperation. Up to seven attempts were made to contact the household, and once contact was made, up to an additional seven attempts were made to gain cooperation with the selected respondent. Furthermore, because of the short field period, in the last week of interviewing, all numbers that had been dialed seven times without making contact were re-opened to make additional attempts to contact those households. The final contact rate was 70.0 percent (4481 contacted households/ 6397 working numbers).

To improve cooperation rates, interviewers were trained on persuasion techniques, and were asked to be extremely careful with coding cases as refusals, since limited follow-up is conducted for refusals. Specifically, interviewers were asked to code responses such as "Iím too busy" or "Iím not interested" as callbacks instead of refusals, if this comment was made before the actual respondent was selected for the study. Towards the end of the field period, a persuasion letter was mailed to households considered as "soft" refusals (who refused to participate but seemed receptive to persuasion efforts). The letter, signed by Max Larsen, Senior Vice President of Gallup, was mailed on December 19, 2000 to 176 households. The refusal rate on this study was 40.2 percent (1,802 refusals/ 4,481 contacted residential numbers).

Table 1 below shows the final dispositions for the sample. A total of 5,603 of the sample numbers were determined to be unassigned or business numbers. In addition, Gallup interviewers were unable to reach anyone at 1,376 of the numbers to determine their residential status because of repeated busy signals or no answers. Within the remaining numbers classified as residential, the interviewers made contact with 4,481 households and completed 1,658 interviews. Overall, Gallupís conservative estimate of the response rate was 30 percent.

We estimated the response rate using the definition recommended by the Council of American Survey Research Organizations (CASRO). Under this definition, the numerator for the response rate is the number of completed interviews (1,658) and the denominator the estimated number of eligibles (in this case, eligible households) in the sample. We estimate that 47.3 percent of the sample numbers were working residential numbers (= 5,021 residential numbers /(12,000 total numbers Ė 1,376 non-contacted numbers). In addition, we estimate that 97.9 percent of these residences included one or more eligible members (1,679 eligible/1,714 screened). Overall, then, we estimate that the sample numbers were linked to 5,557 eligible households (12,000 x .473 x .979); this yields a response rate of 29.8 percent (1,658 / 5,557).

Table 1. Disposition of Sample Cases

Disposition

Frequency

All Sample Numbers 12,000
Prescreened as unassigned by SSI 2,110
Numbers Dialed 9,890
Business Numbers 1,330
Unassigned Numbers 2,163
No Answer/Busy 1,376
Answering Machine (Residential) 540
Other Residential Numbers 4,481
Contacted Residential Numbers 4,481
Not Screened 2,767
Refusal/Breakoff 1,802
Other (Language problems etc.) 332
Callback 633
Screened 1,714
Ineligible 35
Complete 1,658
Callback 21

Weighting

Weights are applied to survey data for a variety of reasons. Often samples are selected with unequal probabilities, and the weights compensate for these differences in selection probabilities. In addition, the weights may attempt to adjust for the effects of nonresponse and under-coverage.

The first weights for this study were base weights, or probability weights (w1) that reflected the respondentís selection probability:

Graphical respresentation of the formula using mathematical symbols. Formula - Weight One equals the quantity one divided by the selection probability times the quantity one divided by the number of household telephone lines.

where ph represents the selection probability for the sample numbers within each of the two sample strata h; and tjh the number of telephone lines linked to the respondentís household. To reduce the variability in the base weights, we set the value of tjh to 9 for those cases (n=3) who reported more than nine telephone lines. Cases where number of phone lines was not available, or where the number of phone lines given by the respondent was implausibly high (97 or more) were edited to 1. The sampling probability was calculated by dividing the sample size by the frame totals as provided by SSI for each of the two income strata.

A normalized base weight was then produced by dividing each base weight by the mean of the weights, so that the sum of the normalized weights is equal to the sample size. In applications where population projections are not needed, or when weights cannot be reliably projected to the target population, normalized weights are generally preferred.

The final weights were produced by adjusting the base weights for nonresponse. Gallup and FDIC chose three sampling variables, Census region, urbanicity, and the income sampling strata to use as categories for the adjustment cells, yielding 24 cells. Because there were relatively few high-income rural cases, the four rural high-income cells were collapsed into one, yielding a total of 21 cells.

Within each adjustment cell Gallup computed the eligibility rate based on cases which were screened out during interviewing. This rate was then used to compute a response rate in each cell, using a CASRO-type formula. The nonresponse-adjusted weights (w2) were then produced as follows:

Graphical respresentation of the formula using mathematical symbols. Formula - Weight Two equals Weight One multiplied by one divided by the response rate adjustment.

where rk is the response rate for adjustment cell k. Finally the weights were normalized as described above. Table 2 contains the non-response adjustment factors by cell.

The reader should be aware that the nonresponse adjustment does not yield weights that project to the number of eligible households. Although it would be possible to attempt this by post-stratifying to adjusted Census population counts, Gallup does not believe that reliable estimates of the target population can be constructed. For this reason (as well as the reasons described above) Gallup recommends that the normalized weights be used whenever weighted estimates are desired.

Table 2 Ė Non-response Adjustment Factors

INCOME

CENSUS REGION

URBAN

ADJ

HIGH

NORTHEAST

SUBURBAN

5.94054

HIGH

NORTHEAST

URBAN

8.625

HIGH

MIDWEST

SUBURBAN

5.2454

HIGH

MIDWEST

URBAN

5.33333

HIGH

SOUTH

SUBURBAN

4.93082

HIGH

SOUTH

URBAN

6.28947

HIGH

WEST

SUBURBAN

6.72727

HIGH

WEST

URBAN

8.95

HIGH

ALL

RURAL

3.84615

LOW

NORTHEAST

RURAL

6.55556

LOW

NORTHEAST

SUBURBAN

5.03226

LOW

NORTHEAST

URBAN

8.33333

LOW

MIDWEST

RURAL

4.67416

LOW

MIDWEST

SUBURBAN

4.60638

LOW

MIDWEST

URBAN

6.12698

LOW

SOUTH

RURAL

4.69369

LOW

SOUTH

SUBURBAN

5.70064

LOW

SOUTH

URBAN

6.80808

LOW

WEST

RURAL

7.56

LOW

WEST

SUBURBAN

6.0119

 

Imputation

  1. General Guidelines
  2. There were two methods used to edit and impute the data for the FDIC Insurance Study. A group of opinion questions where the respondent refused to answer were changed to Donít Know. After editing, a group of factual items where the respondents refused to answer or said they didnít know were imputed using a hotdeck. Both procedures are described in detail below.

  3. Editing
  4. There was one edit made to the data that was not flagged, in order to make some respondentsí answers logically consistent. If a respondent indicated in QN25 that they currently had more than $100,000 of total deposits in one or more banks, then QN27 was edited to be consistent. (QN27 asked if the respondent had ever held more than $100,000 in banks.) There were 77 cases that were changed as a result.

    Additionally, there were several opinion questions where respondents refused to give an answer. Personal opinions were treated differently than factual questions for purposes of imputation. For opinion items, it was assumed that refusal was equivalent to lack of any knowledge or opinion on the subject. They were therefore coded as DK, and flagged appropriately. There was also one question where both DK and refusal were coded as Other, and one question where both DK and refusal were coded as None.

    It should be noted that DK is considered a valid and important response to opinion questions, and therefore these responses were not imputed. A list of the questions and which values were altered can be found in Table 3 below.

  5. Hotdeck Imputation
  6. Factual items where the respondent either answered DK or refused were candidates for hotdeck imputation. The specific method used was a nearest neighbor hotdeck, where groups of missing items were imputed from a single donor by sorting the cases using a list of covariates, then filling in responses from the nearest complete case.

    Covariates used for sorting were chosen using an automated search of bivariate correlations. For each variable a list of the ten largest correlations (in absolute value) was generated. In order to make use of only relatively strong correlations and to make the data processing more efficient, the number of variables used to sort the hotdeck was limited to three. Because of its strong correlation with many questions, a constructed household savings variable was used in all hotdecks, leaving two additional variables that were chosen on a per variable basis. When groups of variables were imputed together, Gallup tried to pick sorting variables that had high correlations with all variables in the group.

    Question groups and their order were chosen with several goals in mind. The first was to include as many related questions together in the same group as possible. The second was to follow the order of the questionnaire whenever possible. (One exception to this rule was QND7, which was imputed first because of its usefulness as a covariate for later imputation.) The third goal was to impute items that were missing as a result of an imputed response to a filter question. For example, QN29 (Do you currently have any deposits that are uninsured?) and QN30 (If yes to QN29, approximately how much do you currently have in your accounts that is beyond the insurance limit?).

    Overall, 461 cases had at least one value imputed by hotdeck (27.8%.) The remaining 1,197 cases were eligible hotdeck donors and constitute the entire donor pool. A table of question groups listed in the order in which they were imputed, along with the sorting variables used, can be found in Table 4. Also included below is a report of the number of valid responses and imputed cases for each question (Table 5) and a distribution of the total number of times each case was a donor for hotdeck imputation (Table 6.)

Table 3 Ė Response Edits

Question

Original Response

Edited Value

QN2C

REF

DK

QN3

REF

DK

QN4

REF

DK

QN6

REF

DK

QN7C

REF

DK

QN9

REF

DK

QN11A

REF

DK

QN11B

REF

DK

QN13_1

DK, REF

OTHER

QN15

REF

DK

QN17

REF

DK

QN19

REF

DK

QN20

REF

DK/No opinion

QN21

REF

DK

QN22

REF

DK

QN23

REF

DK

QN28E

DK, REF

None/No other reason

QN32

REF

DK

QN33

REF

DK

Table 4 Ė Hotdeck Imputation

Question Group

Sort Variables

QND7

SAVINGS QN14 QND4

QN10

SAVINGS QND7 QND4

QN11A QN11B QN11C QN12 QN13_1 QN13_2 QN13_3 QN14 QN24

SAVINGS QN8A QN2C

QN25 QN27

SAVINGS QND7 QN5

QN26 QN29

SAVINGS QND7 QN5

QN28A QN28B QN28C QN28D QN28E

SAVINGS QN2A QN2B

QN31 QN33

SAVINGS QN25 QN10

QN29 (Note: the original version of this question was preserved.)

SAVINGS QND7 QN5

QN30

SAVINGS QND7 QN5

QN32

SAVINGS QN25 QN10

QN34

SAVINGS QN25 QN10

QN35

SAVINGS QN25 QN10

QND3 QND4 QND5 QND6

SAVINGS QND7 QND1

SAVINGS (Note: constructed savings variable.)

QND7 QND3 SAVE2

Table 5 Ė Number of cases imputed for each question

Variable

Total Responses

Valid Responses

Edited

Hotdeck

QN2A

1658

1658

0

0

QN2B

1658

1658

0

0

QN2C

1658

1656

2

0

QN2D

1658

1658

0

0

QN2E

1658

1658

0

0

QN2F

1658

1658

0

0

QN2G

1658

1658

0

0

QN3

1658

1656

2

0

QN4

1025

1023

2

0

QN5

1658

1658

0

0

QN6

1658

1654

4

0

QN7A

922

922

0

0

QN7B

922

922

0

0

QN7C

922

921

1

0

QN7D

922

922

0

0

QN8A

564

564

0

0

QN8B

564

564

0

0

QN8C

564

564

0

0

QN9

1658

1652

6

0

QN10

1658

1603

0

55

QN11A

1163

1124

1

38

QN11B

1163

1124

1

38

QN11C

1163

1125

0

38

QN12

1163

1111

0

52

QN13_1

393

351

4

38

QN13_2

389

351

0

38

QN13_3

65

27

0

38

QN14

8

-30

0

38

QN15

1658

1649

9

0

QN16

660

660

0

0

QN17

1658

1653

5

0

QN18

652

652

0

0

QN19

1658

1657

1

0

QN20

1658

1655

3

0

QN21

846

845

1

0

QN22

1658

1653

5

0

QN23

1658

1653

5

0

QN24

1163

1068

0

95

QN25

1658

1500

0

158

QN26

238

169

0

69

QN27

1658

1540

0

118

QN28A

511

455

0

56

QN28B

512

460

0

52

QN28C

512

460

0

52

QN28D

512

458

0

54

QN28E

512

452

12

48

QN29

238

203

0

35

QN30

86

61

0

25

QN31

1658

1578

0

80

QN32

498

451

11

36

QN33

1658

1619

0

39

QN34

795

779

0

16

QN35

545

534

0

11

QND1

1658

1658

0

0

QND2

1658

1658

0

0

QND3

1658

1583

0

75

QND4

1658

1628

0

30

QND5

1658

1627

0

31

QND6

1658

1606

0

52

QND7

1658

1395

0

263

QND8

1658

1429

0

229

QND9

1011

962

0

49

QND10

669

645

0

24

QND11

344

336

0

8

QN29NODK

238

226

0

12

Table 6 Ė Frequency of hotdeck donations

Number of donations

Frequency

Percent

0

980

59.11

1

433

26.12

2

120

7.24

3

42

2.53

4

25

1.51

5

25

1.51

6

11

0.66

7

3

0.18

8

6

0.36

9

3

0.18

10

3

0.18

11

2

0.12

13

1

0.06

14

1

0.06

15

1

0.06

18

1

0.06

32

1

0.06

Total

1658

100



Last Updated 04/26/2001 supervision@fdic.gov