Skip to main content
Log in

Estimating Cash Usage: The Impact of Survey Design on Research Outcomes

  • Published:
De Economist Aims and scope Submit manuscript

Abstract

We employ a unique dataset of transaction records to analyse the impact of survey set-up on consumers’ payments registration behaviour. Survey data are used for econometric analyses and validated against other payments data. The results reveal that the length of the registration period influences consumers’ registration of payments. Measurement errors are minimised when consumers use a self-reported transaction diary for one single day. Around 40 % of the transactions registered in a one-day survey are missed out in a one-week survey. Apart from payments research, the results are, among others, also relevant for household expenditure and marketing research.

This is a preview of subscription content, log in via an institution to check access.

Access this article

Price excludes VAT (USA)
Tax calculation will be finalised during checkout.

Instant access to the full article PDF.

Institutional subscriptions

Fig. 1
Fig. 2

Similar content being viewed by others

Notes

  1. The panels are continuously monitored and adjusted to ensure representativeness on a number of dimensions, such as on age, gender, education and region. For both panels, new members are mainly recruited in a traditional way, i.e. by phone or letter. In order to prevent overrepresentation of heavy internet users, free online registering is hardly used. Moreover, in order to maximise response, panel members are approached for survey participation not more than a few times a year.

  2. Overall, the weighting of the data reduced the estimated number of payments made. This holds for both cash and debit card payments, as well as for all seven studies. Yet, the five day surveys (Study 1–5) were more heavily adjusted than the two week surveys (Study 6 and 7). The main reason lies within the correction made on the number of ‘zero payments’, as the non-response check showed that the share of ‘non-paying’ non-responders was highest among the five day surveys. This in itself is not surprising, as the likelihood of not making a transaction during one particular day is much higher than the likelihood of not making any during one entire week.

  3. In case of Study 6 and Study 7, the average number of recorded payments across the seven days is presented.

  4. The large retail chains retrieved transaction information from their databases. Some small and medium sized retailers also provided information from their databases, others made a best guess estimate. Since no information is provided on the standard deviations in these transaction records, we were not able to take them into account in the validation tests by, for instance, employing two sample \(t\)-tests. The tests we performed are therefore more conservative. However, we do not expect this to affect the final results substantially, as the differences between retailer and consumer data are often large and the \(p\)-values of the equality tests are well below the critical value of 0.05. We expect that the number of cash payments reported in Study 5 may have turned out not too differ significantly from the retailer data.

  5. If the sample variance is more than twice the sample mean, data are likely to be overdispersed, even after inclusion of regressors. A sound practice is therefore to estimate both Poisson and negative binomial models.

  6. The results of the validation exercise in Sect. 4 revealed that debit card payments are not under- but overreported. A zero deflated model may in that case be more suitable. However, this method does not account for the heterogeneous character of people’s payment preferences. It is interesting for future research to further explore the possibilities of zero deflated models and finite mixed models.

  7. There is a variation in this variable in all seven studies, also in the studies where participants participated online. That is, the online respondents did not necessarily need to have access to the internet at home, as they were also able to participate when being at school or at work (which was the case for about 10 % of the respondents). Overall, due to the high internet penetration in the Netherlands, the variations on this variable are small, from 94 % in Study 5 to 99 % in Studies 6 and 7.

  8. Due to the diary structure, there are interdependencies between the reported payments of respondents participating in studies 6 and 7, as they reported their payments for seven days in a row. We took this into account by clustering the standard errors by respondents, using the individual respondent as a cluster. For background information on clustering, see Wooldridge (2003) or Cameron and Miller (2011).

  9. The regression coefficients referring to the ‘always zero payments’ part of the ZINB model can be interpreted in the same way as coefficients from the binomial logistic model. The coefficients referring to the true number of payments are to be interpreted as coefficients for a negative binomial model, with the expected number of payments changing with exp(coef) for each unit increase in the corresponding explanatory variable. Note that the ZINB model is known for estimation problems of the ‘always zero payments’ part. Unfortunately, some of the estimated coefficients in our model suffered from this drawback (i.e. the coefficient \(-\)13.39 for primary education). However, the coefficients referring to the impact of the study design remained unaffected and turned out to be fairly robust against alternative model specifications. Therefore, our main conclusion with respect to the research questions remains unaffected.

  10. In order to check the robustness of the estimation results, we re-run the main regressions without including the demographic variables. Overall, there were no substantial changes to our conclusions on the role of study design; the estimation results regarding the impact of the ‘study’ variables turned out to be broadly similar thus very robust, both with respect to the magnitude of the estimated effects, as well as their significance.

  11. We did not find any significant differences for individual transaction categories, whereas the results did differ on an overall level.

  12. We also performed tests on the data from the period after the reminder only. The results remained the same, however.

  13. It may be interesting to examine registration periods of 2 or 3 days. Longer registration periods have the advantage of collecting more information at lower costs and may reduce part of the social desirability error that one-day surveys suffer from.

  14. Source: Eurostat Statistics 2009, available at: http://epp.eurostat.ec.europa.eu/cache/ITY_OFFPUB/KS-SF-11-066/EN/KS-SF-11-066-EN.PDF.

References

  • Ahmed, N., Brzozowski, M., & Crossley, T. F. (2006). Measurement errors in recall food consumption data, The Institute for Fiscal Studies Working Paper 06/21. Cambridge.

  • Alessie, R., Gradus, R., & Melenberg, B. (1990). The problem of not observing small expenditures in a consumer expenditure survey. Journal of Applied Econometrics, 5, 151–166.

    Article  Google Scholar 

  • Battistin, E., Miniaci, R., & Weber, G. (2003). What do we learn from recall consumption data? The Journal of Human Resources, 38, 354–385.

    Article  Google Scholar 

  • Bergman, M., Guibourg, G., & Segendorf, B. (2007). The costs of paying-private and social costs of cash and card payments, Sveriges Riksbank Working Paper Series No. 212, Stockholm.

  • Bethlehem, J. G. (2008). Representativity of Web Surveys—an Illusion? In Stoop, I. & Wittenberg, M. (Eds.), Access panels and online research, panacea or pitfall? Proceedings of the DANS symposium (pp. 19–44). Amsterdam, Aksant.

  • Boeschoten, W. C. (1992). Currency use and payment patterns, PhD thesis, Amsterdam.

  • Borzekowski, R., Kiser, E. K., & Ahmed, S. (2008). Consumers’ use of debit cards: Patterns, preferences, and price response. Journal of Money, Credit and Banking, 40, 149–172.

    Article  Google Scholar 

  • Bound, J., Brown, C., & Mathiowetz, N. (2001). Measurement error in survey data. In J. J. Heckman & E. Leamer (Eds.), Handbook of Econometrics (pp. 3705–3843). Amsterdam: Elsevier Science.

  • Bounie, D. & François, A. (2006). Cash, check or bank card? The effects of transaction characteristics on the use of payment instruments, Telecom Paris Economics and Social Sciences Working Paper No. ESS-06-05, Paris.

  • Bowling, A. (2005). Mode of questionnaire administration can have serious effects on data quality. Journal of Public Health, 27, 281–291.

    Article  Google Scholar 

  • Brits, J. H., & Winder, C. C. A. (2005). Payments are no free lunch. Amsterdam: DNB Occasional Studies.

    Google Scholar 

  • Brown, N. R., Rips, L. J., & Shevell, S. K. (1985). The subjective dates of natural events in very-long-term memory. Cognitive Psychology, 17, 139–177.

    Article  Google Scholar 

  • Cameron, A. C., & Trivedi, P. K. (1998) Regression analysis of count data. Econometric Society Monographs no. 30, Cambridge.

  • Cameron, A. C., & Miller, D. L. (2011). Robust inference with clustered data. In A. Ullah & A. D. Giles (Eds.), Handbook of empirical economics and finance (pp. 1–28). New York: CRC Press.

    Google Scholar 

  • Chakravorti, S. (2006). The evolving payments landscape and its implications for monetary policy. In S. W. Schmitz & G. Wood (Eds.), Institutional change in the payments system and monetary policy (pp. 81–92). London: Routledge.

  • Currence (2005–2007). Monitor Consumptieve Toonbankbetalingen in Nederland, Final Report, Amsterdam.

  • EIM (2007) Het Toonbankbetalingsverkeer in Nederland: Kosten en Opbrengsten van Toonbankinstellingen in Kaart Gebracht. Zoetermeer: Final Report.

  • Estelami, H., & Lehmann, D. R. (2001). The impact of research design on consumer price recall accuracy: An integrative review. Journal of the Academy of Marketing Science, 29, 29–36.

    Article  Google Scholar 

  • Freedman, C. (2000). Monetary policy implementation: Past, present and future—Will the advent of electronic money lead to the demise of central banking? International Finance, 3, 211–227.

    Article  Google Scholar 

  • Gibson, J. (2002). Why does the engel method work? Food demand, economies of size and household survey methods. Oxford Bulletin of Economics and Statistics, 64, 341–360.

    Article  Google Scholar 

  • Gibson, J., & Kim, B. (2007). Measurement error in recall surveys and the relationship between household size and food demand. American Journal of Agricultural Economics, 89, 473–489.

    Article  Google Scholar 

  • Holbrook, A. L., & Krosnick, J. A. (2010). Social desirability bias in voter turnout reports: Tests using the item count technique. Public Opinion Quarterly, 74, 37–67.

    Article  Google Scholar 

  • Humphrey, D. B., Pulley, L. B., & Vesala, J. M. (2000). The check’s in the mail: Why the United States lags in the adoption of cost-saving electronic payments. Journal of Financial Services Research, 17, 17–39.

    Article  Google Scholar 

  • Kreuter, F., Presser, S., & Tourangeau, R. (2008). Social desirability bias in CATI, IVR and web surveys. Public Opinion Quarterly, 72, 847–865.

    Article  Google Scholar 

  • Krosnick, J. A., & Alwin, D. F. (1987). An evaluation of a cognitive theory of response-order effects in survey measurement. Public Opinion Quarterly, 51, 201–219.

    Article  Google Scholar 

  • Linton, M. (1982). Transformations of memory in everyday life. In U. Neisser (Ed.), Memory observed (pp. 77–91). New York: W.H. Freeman and Co.

    Google Scholar 

  • Lynn, P., Jäckle, A., Jenkins, S. P., & Sala, E. (2004). The impact of interviewing method on measurement error in panel survey measures of benefit receipt: Evidence from a validation study, ISER Working Papers No. 2004–28, Colchester.

  • Scott, K., & Okrasa, W. (1998). Analysis of Latvia diary experiment. Washington DC: World Bank Development Research Group Research Paper.

  • Sudman, S., & Lannom, L. (1980). Health care surveys using diaries, DHHS Publication no. PHS 80–3279, Washington, DC.

  • Sudman, S., & Bradburn, N. (1973). Effects of time and memory factors on response in surveys. Journal of the American Statistical Association, 68, 805–815.

    Article  Google Scholar 

  • Tincello, D. G., Williams, K. S., Joshi, M., Assassa, R. P., & Abrams, K. R. (2007). Urinary diaries. Obstetrics and Gyneacology, 109, 227–280.

    Google Scholar 

  • TNS NIPO (2003). Contante transacties in Nederland. Amsterdam: Final Report.

  • Van Praag, B. M. S., & Vermeulen, E. M. (1993). A count-amount model with endogeneous recording of observations. Journal of Applied Econometrics, 8, 383–395.

    Article  Google Scholar 

  • Von Kalckreuth, U., Schmidt, T., & Stix, H. (2009). Choosing and using payment instruments: Evidence from German micro-data. Frankfurt: ECB Working Paper.

  • Wooldridge, J. M. (2003). Cluster-sample methods in applied econometrics. American Economic Review, 93, 133–138.

    Article  Google Scholar 

  • Wutich, A. (2009). Estimating household water use: A comparison of diary, prompted recall and free recall methods. Field Methods, 21, 49–68.

    Article  Google Scholar 

Download references

Author information

Authors and Affiliations

Authors

Corresponding author

Correspondence to Anneke Kosse.

Rights and permissions

Reprints and permissions

About this article

Cite this article

Jonker, N., Kosse, A. Estimating Cash Usage: The Impact of Survey Design on Research Outcomes. De Economist 161, 19–44 (2013). https://doi.org/10.1007/s10645-012-9200-2

Download citation

  • Published:

  • Issue Date:

  • DOI: https://doi.org/10.1007/s10645-012-9200-2

Keywords

JEL codes

Navigation