Athey, S, C. Catalini and C. Tucker. 2017. “The Digital Privacy Paradox: Small Money, Small Costs, Small talk.†National Bureau of Economic Research Working Paper No. 23488.
Chang, T. and P. S. Kott. 2008. “Using Calibration Weighting to Adjust for Nonresponse Under a Plausible Model.†Biometrika 95 (3): 555–571.
- Chart 1: Price and number of Bitcoin transactions, 2012–19 (monthly average) Wave 1/2 Wave 3 Wave 4 0 5 10 15 20 25 30 0 50 100 150 200 250 300 350 400 2012 2014 2016 2018 2013 2015 2017 2019 Transactions Price d
Paper not yet in RePEc: Add citation now
- Chart 2: Use of Bitcoin, 2017–18 (a) Buying goods and services 52% 16% 33% 41% 13% 46% 0 20 40 60 80 100 Rarely Sometimes Often 2017 2018 d
Paper not yet in RePEc: Add citation now
- Chart 3: Current and past ownership of Bitcoin, 2016–18 3% 2% 4% 1% 5% 3% 0 2 4 6 8
Paper not yet in RePEc: Add citation now
- Chart 4: Bitcoin holdings, 2016–18 54% 39% 7% 68% 19% 13% 87% 9% 5% 0 20 40 60 80 100 Less than 1 1?10 More than 10 2016 2017 2018 hj Percent Note: In 2018, we asked respondents to report their holdings as a continuous range, denominated in Canadian dollars. For comparability across years, in 2018 we used the prevailing price when the survey was conducted to denominate respondents’ holdings in Bitcoin. The sample consists of 99 Canadians aged 18 or older who reported they owned Bitcoin in 2018; similarly, 117 in 2017, and 58 in 2016. All estimates were calculated using MICAL (multiple imputation in calibration) survey weights.
Paper not yet in RePEc: Add citation now
- Chart 5: Ownership estimates by weighting method, 2016–18 2% 3% 4% 5% 6% 7% 8% 2016 2017 2018 Year PS Cal3 Cal4 Cal5 Cal7 MICAL7 Raw Note: This figure compares the different calibration methods listed in Section A.2.5. We plot estimates of Bitcoin ownership for each methodology and each year of the survey. Our final estimates are based on MICAL7. The sample consists of 1,997 respondents in 2016, 2,623 respondents in 2017 and 1,987 respondents in 2018.
Paper not yet in RePEc: Add citation now
Chen, H. and Q. R. Shen. 2015. “Variance Estimation for Survey-Weighted Data Using Bootstrap Resampling Methods: 2013 Methods-of-Payment Survey Questionnaire.†Bank of Canada Technical Report No. 104.
Chen, H., M. H. Felt and C. S. Henry. 2018. “2017 Methods-of-Payment Survey: Sample Calibration and Variance Estimation.†Bank of Canada Technical Report No. 114.
Cosslett, S. R. 1981. “Maximum Likelihood Estimator for Choice-Based Samples.†Econometrica 49 (5): 1289–1316.
- Current owners Past owners 2016 2017 2018 d
Paper not yet in RePEc: Add citation now
- Deville, J. C., C. E. Särndal and O. Sautory. 1993. “Generalized Raking Procedures in Survey Sampling.†Journal of the American Statistical Association 88 (423): 1013–1020.
Paper not yet in RePEc: Add citation now
- Ernst and Young. 2019. Fifth Report of the Monitor. Bankruptcy Proceedings in the Supreme Court of Nova Scotia, Hfx No. 484742.
Paper not yet in RePEc: Add citation now
- Fagerland, M. W. and D. W. Hosmer. 2013. “A Goodness-of-Fit Test for the Proportional Odds Regression Model.†Statistics in Medicine 32 (13): 2235–2249.
Paper not yet in RePEc: Add citation now
- Fagerland, M. W., D. W. Hosmer and A. M. Bofin. 2008. “Multinomial Goodness-of-Fit Tests for Logistic Regression Models.†Statistics in medicine 27 (21): 4238–4253.
Paper not yet in RePEc: Add citation now
- Financial Conduct Authority. 2019. Cryptoassets: Ownership and Attitudes in the UK; Consumer Survey Research Report. March.
Paper not yet in RePEc: Add citation now
- For 2018, we updated the methodology using newly available 2016 census data to account for the other demographic variables identified above. Aiming for consistency with the 2017 Methods-of-Payment Survey (MOP), we follow the spirit of Chen, Felt and Henry (2018) in using raking (Deville, Särndal and Sautory 1993) to match our survey to 2016 population totals. In general, for respondent i we calculate weights wi given by (1) wi = wbiwnra,iwcal,i, where wbi is the base weight (or design weight in a probability survey), wnra,i is a nonresponse adjustment (NRA) that scales the sample to the panel it is drawn from, and wcal,i is a “calibration†that scales the panel to the population. We make some simplifications from Chen, Felt and Henry (2018). First, we choose base weight wbi = 1 rather than poststratification.
Paper not yet in RePEc: Add citation now
Henry, C. S., K. P. Huynh and G. Nicholls. 2018. “Bitcoin Awareness and Usage in Canada.†Journal of Digital Banking 2 (4): 311–337.
Henry, C. S., K. P. Huynh and G. Nicholls. 2019. “Bitcoin Awareness and Usage in Canada: An Update.†The Journal of Investing 28 (3): 21–31.
- Hundtofte, S., M. Lee, A. Martin and R. Orchinik. 2019. “Deciphering Americans’ Views on Cryptocurrencies.†Liberty Street Economics (blog), Federal Reserve Bank of New York.
Paper not yet in RePEc: Add citation now
Lusardi, A and O. S. Mitchell. 2011. “Financial Literacy and Planning: Implications for Retirement Wellbeing.†National Bureau of Economic Research Working Paper No. 17078.
Lusardi, A. and O. S. Mitchell. 2014. “The Economic Importance of Financial Literacy: Theory and Evidence.†Journal of Economic Literature 52 (1): 5–44.
- Lusardi, A., C. B. Scheresberg and M. Avery. 2018. Millennial Mobile Payment Users: A look into their Personal Finances and Financial Behaviors. GFLEC Insights Report, GFLEC.
Paper not yet in RePEc: Add citation now
- Marchenko, Y. V. and W. Eddings. 2011. “A Note on How to Perform Multiple-Imputation Diagnostics in Stata.†College Station: StataCorp.
Paper not yet in RePEc: Add citation now
- Murray, J. S. 2018. “Multiple Imputation: A Review of Practical and Theoretical Findings.†Statistical Science 33 (2): 142–159.
Paper not yet in RePEc: Add citation now
- Ontario Securities Commission. 2018. Taking Caution: Financial Consumers and the Cryptoasset Sector. Technical Report.
Paper not yet in RePEc: Add citation now
- Percent (b) Making person-to-person transfers 58% 11% 31% 48% 12% 40% 0 20 40 60 80 100 Rarely Sometimes Often 2017 2018 d Percent Note: The “Rarely†category consists of Canadians who used Bitcoin at most once a year for transactions.
Paper not yet in RePEc: Add citation now
- Percent Note: The sample includes 99 Canadians aged 18 or older who reported they owned Bitcoin in 2018; similarly, 117 in 2017, and 58 in 2016. Additionally, the sample includes 45 past owners in 2018, as well as 37 in 2017, and 41 in 2016. All estimates were calculated using MICAL (multiple imputation in calibration) survey weights.
Paper not yet in RePEc: Add citation now
- Remaining methodologies (Cal4, Cal5, MICAL7) produce more moderate estimates of ownership in 2018 and are fairly similar across years. Taking MICAL7 as the “least biased,†we see that ownership is adjusted upward from the raw sample in 2016, downward slightly in 2017, and upward slightly in 2018. This again reflects the changing demographic trends for Bitcoin ownership and illustrates the need to adjust for as many demographics as possible to reduce the bias in our non-probability sample.
Paper not yet in RePEc: Add citation now
- Rubin, D. B. 2004. Multiple Imputation for Nonresponse in Surveys. Volume 81. John Wiley & Sons.
Paper not yet in RePEc: Add citation now
- Second, we cannot perform NRA due to a lack of data on non-respondents, so set wnra,i = 1. Finally, we avoid interaction terms in our calibration model. Perhaps as a result of these simplifications, we find fewer and less severe outlying weights, and so do not perform weight trimming. We thus proceed with calibration weights wi = wcal,i, which we will compute using the raking ratio method and population counts taken from the 2016 census. Note that this should not be interpreted to mean that non-response is unaccounted for. Calibration is known to adjust for non-response and under-coverage for certain model assumptions (Chang and Kott 2008).
Paper not yet in RePEc: Add citation now
- Statistics Canada. 2018. Digital Economy, July 2017 to June 2018. Technical Report.
Paper not yet in RePEc: Add citation now
Stix, H. 2019. “Ownership and Purchase Intention of Crypto-Assets – Survey Results.†Oesterreichische Nationalbank Working Paper No. 226.
- Table 1: Bitcoin knowledge questions Question Response options The total supply of Bitcoin is fixed. True False Bitcoin is backed by a government. True False All Bitcoin transactions are recorded on a True distributed ledger that is publicly accessible. False Note: This table shows the three Bitcoin knowledge questions, which were also asked in the 2017 BTCOS. The correct answers are highlighted in bold.
Paper not yet in RePEc: Add citation now
- Table 14: Goodness-of-fit tests for imputation model Employment status (multinomial logit) Baseline 2016 2017 2018 Combined Employed 0.333 0.106 0.064 0.000 Unemployed 0.521 0.199 0.593 0.150 Not in labour force 0.073 0.199 0.003 0.000 Income (ordered logit) 2016 2017 2018 Combined 0.482 0.173 0.108 0.014 Note: This table reports p-values from goodness-of-fit tests for imputation models of employment and income. These are based on the Hosmer-Lemeshow test for multinomial logit (Fagerland, Hosmer and Bofin 2008) and ordinal logit (Fagerland and Hosmer 2013), respectively. For employment, we report the tests for different baselines, which is not an issue for the ordered logit.
Paper not yet in RePEc: Add citation now
- Table 3: Financial literacy and Bitcoin knowledge scores Financial literacy Bitcoin knowledge 2018 2017 2018 Overall Adopters Overall Adopters Overall Adopters Low 27 38 55 24 61 19 Medium 36 33 38 49 33 52 High 37 29 6 27 6 29 Note: This table reports the share of Canadians, in percent, in each category of financial literacy or Bitcoin knowledge. The sample consists of 99 adopters in 2018 and 117 in 2017. All estimates were calculated using MICAL (multiple imputation in calibration) survey weights.
Paper not yet in RePEc: Add citation now
- Table 4: Awareness of Bitcoin in Canada, 2016–18 2016 2017 2018 Overall 62 83 89 Gender Male 71 90 93 Female 54 77 85 Age 18–34 69 87 91 35–54 58 82 88 55+ 62 82 88 Education High school 55 76 84 College 59 85 90 University 78 92 95 Income ($) <30,000 49 74 87 30,000–69,999 61 82 88 70,000+ 69 87 91 Region British Columbia 74 93 94 Prairies 66 84 89 Ontario 64 85 92 Quebec 49 75 84 Atlantic 65 80 83 Financial literacy Low . . 80 Medium . . 90 High . . 94 Note: This table reports the percentage of Canadians who were aware of Bitcoin in 2016, 2017 and 2018.
Paper not yet in RePEc: Add citation now
- Table 8: Preferences for online transactions, Canadians vs. adopters Overall Canadians Bitcoin adopters Privacy Security Acceptance Ease Privacy Security Acceptance Ease Most 14 61 11 15 26 28 20 26 More 39 19 20 22 26 28 20 26 Less 22 12 29 37 22 26 29 23 Least 25 8 40 26 26 18 31 25 Note: This table shows the percentage of Canadians who ranked each feature of online transactions, from most to least important. The estimates in each column sum vertically to 100 percent. The sample consists of 99 adopters in 2018, 117 in 2017 and 58 in 2016. All estimates were calculated using MICAL (multiple imputation in calibration) survey weights.
Paper not yet in RePEc: Add citation now
- Table 9: Cash management and Bitcoin adoption, 2017–18 No cash Already Plans to go cashless Median ($) on hand (%) cashless (%) within 5 years (%) Overall 2017 MOP 40 9 . . 2017 BTCOS 40 8 . . 2018 BTCOS 40 8 7 5 Adopters 2017 MOP (4.0%) 65 8 . . 2017 BTCOS (4.3%) 100 4 . . 2018 BTCOS (5.2%) 200 8 18 17 Note: We report results from three surveys conducted by the Bank of Canada: the 2017 BTCOS, the 2018 BTCOS and the 2017 Methods-of-Payment (MOP) Survey. The sample consists of 99 adopters in 2018 and 117 in 2017. All BTCOS estimates were calculated using MICAL (multiple imputation in calibration) survey weights and the 2017 MOP estimates used survey weights as well.
Paper not yet in RePEc: Add citation now
- Technology Strategies International. 2019. Canadian Payments Forecast 2019. Technical Report.
Paper not yet in RePEc: Add citation now
- The “Sometimes†category constitutes those who used Bitcoin between a few times a year to once a month, and “Often†constitutes those who used Bitcoin at least a few times a month for transactions. The sample consists of 99 Canadians aged 18 or older who reported they owned Bitcoin in 2018; and similarly, 117 in 2017. All estimates were calculated using MICAL (multiple imputation in calibration) survey weights.
Paper not yet in RePEc: Add citation now
- The sample consists of 99 adopters in 2018, 117 in 2017 and 58 in 2016. All estimates were calculated using MICAL (multiple imputation in calibration) survey weights.
Paper not yet in RePEc: Add citation now
- To assess the chosen imputation model we adapt the guidelines of Marchenko and Eddings (2011) for use in our context for multinomial and ordinal logit models. First, we run the imputation models for each year using complete cases in the data, and assess their goodness of fit. Second, we examine if there is evidence to reject the assumption of data missing completely at random (MCAR). In addition to their recommendations, we describe our choice of the number of imputations to be used in each bootstrap replication.
Paper not yet in RePEc: Add citation now
- To summarize, this procedure consists of two core parts. First, we must use a calibration model to estimate weights for a given set of demographics. Our selection of variables for this raking procedure is discussed in the previous sections. Second, we must estimate an imputation model. This is complicated by the fact that multiple variables could be missing at once; in our case, we find item non-response in both employment and income. Thus, we use multivariate imputation by chained equations, or MICE (see, e.g., White, Royston and Wood 2011), to generate imputations for both employment and income using other demographic characteristics as predictors. Our algorithm for estimating (Ω) by ̂MICAL(S) is as follows: 1. Using MICE, impute unobserved demographics xu i J times: f xu ij, j = 1, ..., J. 2. For each f xu ij, use raking to produce weights f wij. 3. Compute final weights wi = 1 J PJ j=1 f wij. 4. ̂MICAL(S) = 1 N P i∈S wiyi.
Paper not yet in RePEc: Add citation now
- Transactions (thousands) Price (thousands) Note: This graph shows the price of Bitcoin in Canadian dollars and the number of daily transactions made with Bitcoin, averaged over each month from January 2012 to January 2019. The data series for price starts at March 12, 2013. The green vertical lines show when the first two waves of the BTCOS were in the field, the red vertical line shows the third wave and the blue line indicates the most recent iteration, the 2018 BTCOS. The last monthly observation is July 2019. Sources: Daily Transactions (Charts.Bitcoin.com/BTC); Bitcoin Prices (BTC/CAD) (Yahoo! Finance).
Paper not yet in RePEc: Add citation now
- Van Buuren, S. 2018. Flexible Imputation of Missing Data. Chapman and Hall/CRC.
Paper not yet in RePEc: Add citation now
- White, I. R., P. Royston and A. M. Wood. 2011. “Multiple Imputation Using Chained Equations: Issues and Guidance for Practice.†Statistics in Medicine 30 (4): 377–399.
Paper not yet in RePEc: Add citation now
- X b=1 (̂MICAL(Sb) − ̂MICAL(S))2 . (9) A.2.5 Comparison We now compare estimates of Bitcoin ownership using the following methodologies: • PS: Post-stratification on age, gender and province. This is the methodology used for the first two iterations of the BTCOS. See Henry, Huynh and Nicholls (2018) for more details. • Cal3: Raking on age, gender and province. • Cal4: Raking on Cal3 + education. • Cal5: Raking on Cal4 + marital status. • Cal7: Raking on Cal5 + employment and income, where those with missing values are dropped. • MICAL7: Cal7, but with missing values for employment and income handled by multiple imputation.
Paper not yet in RePEc: Add citation now