Advance Payment Systems Paying Too Much Today and

  • Slides: 24
Download presentation
Advance Payment Systems: Paying Too Much Today and Being Satisfied Tomorrow Forthcoming in International

Advance Payment Systems: Paying Too Much Today and Being Satisfied Tomorrow Forthcoming in International Journal of Research in Marketing, 2015, Vol. 32, Issue 3, 238 -250 Fabian Schulz Goethe University Frankfurt Christian Schlereth WHU – Otto Beisheim School of Management Nina Mazar University of Toronto Bernd Skiera Goethe University Frankfurt

Advance. Ppayment Systems (also referred to as equal billing) Usage prediction for billing cycle

Advance. Ppayment Systems (also referred to as equal billing) Usage prediction for billing cycle Calculation of advance payment rates Determination of actual usage in billing cycle Determination of last billing rate Refund Extra payment 1

Advance Payment Systems (APS) are best known for utility services billing and taxes §

Advance Payment Systems (APS) are best known for utility services billing and taxes § e. g. , electricity, water, gas… taxes But APS are applicable to ANY recurring service where consumption and payments are separated in time Credit cards balances Cloud computing services Pay-as-you-drive car insurances 2

Also outside of Germany, APS are increasingly advertised by electricity service providers Europe US

Also outside of Germany, APS are increasingly advertised by electricity service providers Europe US Advance payment Company France Germany system Optional or offered mandatory EDF Yes Optional ENI Yes Optional GDF Suez Yes Optional Poweo Direct Energy Yes Optional En. BW Yes Mandatory Eon Yes Mandatory EWE Yes Mandatory RWE Yes Mandatory Vattenfall Europe Yes Mandatory Municipale No - Aem No - Edison Sp. A No - Enel No - Hera Group No - EDP Renováveis No - Endesa Yes Optional Acqua Gas Azienda Italy Spain UK Eon Spain No - Gas Natural Yes Optional Iberdrola Yes Optional EDF Energy Yes Optional Eon UK Yes Optional National Grid Yes Optional RWE npower Scottish and Southern nergy Yes Optional 3

Service providers can choose between three payment systems; our focus: advance payments Pros Focus

Service providers can choose between three payment systems; our focus: advance payments Pros Focus of this study Small nonpayment risk Earlier cash flow Low operational costs Customer loyalty Advance payment (Predicted usage paid upfront) Ex ante Prepaid (Usage allowance bought) Payment timing Ex post High Low 4

Do you remember the feeling you had when filing your last tax return? Most

Do you remember the feeling you had when filing your last tax return? Most people have one of the following two reactions: Refund Extra payment 5

Inconsistent research findings on payment sequence preferences Payment sequence preferences for goods Payment sequence

Inconsistent research findings on payment sequence preferences Payment sequence preferences for goods Payment sequence preferences for taxes Income sequence preferences e. g. , Loewenstein & Sicherman (1991) Guyse et al. (2002) Read & Powell (2002) e. g. , Ayers, et al. (1999) Jones (2012) Highfill, Thorson and Weber (1998) e. g. , Prelec and Loewenstein (1998) Patrick and Park (2006) Direction Consumers prefer to Pos-tpay for Pre-pay for hedonic goods utilitarian goods Tax-payers prefer to pre-pay Present value 120. 8 118. 7 Choice 17% 83% Reason Workers prefer rising income streams • Preference to prepay for hedonic goods to enjoy consumption as if it was for free • • • Different results with regards to direction of payment sequence preferences Mainly small experiments in lab (with exception of taxes) Consequences on payment sequence preferences in the consumption sphere is unknown Lack of self-control Asymmetric penalties • • Alignment with productivity Convenience 6

Prospect Theory (e. g. , Silverlining principle) is not able to explain preference for

Prospect Theory (e. g. , Silverlining principle) is not able to explain preference for a refund § Δ: refund (+) or extra payment (-) § b: total yearly bill according to actual consumption 7

Research goals: Analyze payment sequence preferences, as well as causes and consequences Question 1

Research goals: Analyze payment sequence preferences, as well as causes and consequences Question 1 Question Scientific contribution Do preferences for payment sequences exist? Question 2 Do payment sequences have attitudinal consequences? Question 3 Do payment sequences have behavioral consequences? § First paper to examine „irrational behaviour“ in advance payment sequences and whether customers’ preferences shift with relative magnitude of last bill § First paper to examine behavioral and attitudinal consequences § First paper to use survey data and billing data § Support decision for payment system (sdvance vs. pre-payment vs. postpayment) Managerial implications § Support decision for advance payment system design § Provide insights into causes for preferences to support offer design and communication 8

Three research questions three studies Study 1 Study 2 Study 3 Question Do preferences

Three research questions three studies Study 1 Study 2 Study 3 Question Do preferences for payment sequences exist? Do payment sequences have attitudinal consequences? Do payment sequences have behavioral consequences? Type of data Survey 1 Survey 2 merged with billing data 1 Billing data 2 Respondents / customers included General electricity customers Customers of specific European electricity provider incl. churners (2, 672), tariff switchers (3, 411), and passive customers (16, 838) N 259 779 22, 921 9

Three research questions = three studies Study 1 Study 2 Study 3 Question Do

Three research questions = three studies Study 1 Study 2 Study 3 Question Do preferences for payment sequences exist? Do payment sequences have attitudinal consequences? Do payment sequences have behavioural consequences? Type of data Survey 1 Survey 2 merged with billing data 1 Billing data 2 Respondents / customers included General electricity customers Customers of specific European electricity provider N 259 779 22, 921 10

Test for preference of payment sequence preference: Survey 1 Methodology: Survey Versions 1 +

Test for preference of payment sequence preference: Survey 1 Methodology: Survey Versions 1 + 2 Choice experiment set-up: Which sequence would you prefer for expected yearly electricity bill of 600€? Alternative 1: Extra payment sequence Version 1: Equal total payments Alternative 2: Refund sequence Monthly advance payment rate Predicted extra payment at end of year Monthly advance payment rate Predicted refund at end of year Choice-set 1 (low, low) 45€ 60€ 55€ 60€ Choice-set 2 (high, high) 40€ 120€ 60€ 120€ Choice-set 3 (low, high) 45€ 60€ 120€ Choice-set 4 (high, low) 40€ 120€ 55€ 60€ 45€ 60€ 55€ 57. 50€ 40€ 120€ 60€ 115€ 45€ 60€ 115€ 40€ 120€ 55€ 57. 50€ Choice-set 1 (low, low) Version 2: Higher total Choice-set 2 (high, high) payments for refund Choice-set 3 (low, high) sequence Choice-set 4 (high, low) Version 3: Low uncertainty Version 4: High uncertainty 11

Study 1: Percentage of respondents prefering refund sequence Choice-set (Extra Payment, Refund) N Version

Study 1: Percentage of respondents prefering refund sequence Choice-set (Extra Payment, Refund) N Version 1 Version 2 Version 3 Version 4 Total across all versions Equal total payments Higher total payments for refund Low uncertainty High uncertainty 66 60 64 69 259 Choice-set 1 (low, low) 62%** 58% 67%*** 75%*** 65%*** Choice-set 2 (high, high) 64%** 61%* 67%*** 64%*** Choice-set 3 (low, high) 47% 39%* 52% 64%** 50% Choice-set 4 (high, low) 73%*** 67%*** 72%*** 77%*** 72%*** Total across all choice-sets 61%*** 56%** 64%*** 71%*** 63%*** Preference for refund sequences decreases with the relative size of the refund to the extra payment A preference for refund sequences exists Significantly different from 50%: ***p<0. 01; **p<0. 05; *p<0. 1 The majority of respondents is still preferring refund, even if they eventually pay more High uncertainty in usage leads to higher preference for refund sequence 12

Three research questions = three studies Study 1 Study 2 Study 3 Question Do

Three research questions = three studies Study 1 Study 2 Study 3 Question Do preferences for payment sequences exist? Do payment sequences have attitudinal consequences? Do payment sequences have behavioral consequences? Type of data Survey 1 Survey 2 merged with billing data 1 Respondents / customers included General electricity customers Customers of specific European electricity provider N 259 779 Billing data 2 Customers of specific European electricity provider 22, 921 13

Impact of payment sequence on attitudes: billing + survey data Survey data: n=779 (customers

Impact of payment sequence on attitudes: billing + survey data Survey data: n=779 (customers of European electricity provider) Billing data: n=782 (customers of European electricity provider) + § Price awareness Dependent § Likelihood to recommend variables provider Independent variable - Control variables § § Timing January 2012 Gender Age Income Education Price awareness § Refund or extra payment § Relative last billing rate (in % of total yearly rate) § Total electricity spending § Type of contract § Avg. Price / k. Wh Customers received last bill at 16 th of a random month in 2011 14

Simple 2 sample comparison already shows impact of payment sequences on attitudes … Refund

Simple 2 sample comparison already shows impact of payment sequences on attitudes … Refund receivers are half as accurate in their price estimate … … and more likely to recommend their provider Price awareness: Refund receivers vs. % of customers (difference significant at 1% leve)l extra payment makers Average probability to recommend provider on 10 point scale 74, 25% 38. 81% Reading example: People who received a refund with their last billing rate over-/ underestimate their monthly advance payments on average by 38% 6. 44 6. 13 45, 70% 20. 00% Refund receivers Extra payment makers Advance Yearly payments bill Advance Yearly payments. bill Refund receivers Extra payment makers Note: Extra payment makers: N= 384; Refund receivers: N=398; mean difference significant at 5% / 10% confidence level 15

… which is confirmed by linearly regressing relative last billing rate with attitude measures

… which is confirmed by linearly regressing relative last billing rate with attitude measures Results of linear regression models Model 1: Refund Sequence Dummy Model 2: Asymmetric Magnitude Model Advance payment Yearly bill Likelihood of Advance awareness: Absolute percentage recom- mending awareness: percentage provider on Absolute error 10 -point percentage scale error Refund sequence dummy Payment sequence information . 26 *** . 28 ** - - - . 15 . 07 13. 29 *** Relative magnitude of refund Relative magnitude of extra payment sequence Model fit R-square F-Value Number of observations ***p<0. 01; **p<0. 05; *p<0. 1 Note: Control variables not reported due to lack of space 779 payment . 31 * - Likelihood awareness: of recom. Absolute mending percentage provider on error 10 -point scale - - 3. 61 *** 2. 64 *** -. 09 2. 23 *** . 93 . 03 . 15 . 14 . 05 5. 23 *** 2. 60 *** 11. 85 *** 10. 85 *** 3. 41 *** 779 770 Refund sequences reduce price consciousness . 74 ** Yearly bill 779 Refund sequences have a positive influence on likelihood to recommend the provider ( ) 16

Three research questions = three studies Study 1 Study 2 Study 3 Question Do

Three research questions = three studies Study 1 Study 2 Study 3 Question Do preferences for payment sequences exist? Do payment sequences have attitudinal consequences? Do payment sequences have behavioral consequences? Type of data Survey 1 Survey 2 merged with billing data 1 Billing data 2 Respondents / customers included General electricity customers Customers of specific European electricity provider N 259 779 Customers of specific European electricity provider 22, 921 17

Methodology – Question 3: Behavioral consequences We created a sample of churners, tariff switchers,

Methodology – Question 3: Behavioral consequences We created a sample of churners, tariff switchers, and passive customers … All churners of European electricity provider in 2011: N= 2, 672 All tariff switchers of European electricity company in 2011: N= 3, 411 Random sample of passive customers (did not churn, or switch tariffs) in 2011: N= 16, 838 … to calculate impact of payment sequence Questions: § How does type of payment sequence affect odds of churning and tariff switching? § What is effect of magnitude of last billing rate on odds of churning and tariff switching? Methodology: Multinomial logit model: Basis=staying passive § Model 1: Dummy for refund sequence § Model 2: Assymetric magnitude model (absolute of last billing rate/total yearly bill) N= 22, 921 18

Payment sequences have significant impact on behavior More than 50% of churners‘ and tariff

Payment sequences have significant impact on behavior More than 50% of churners‘ and tariff switchers had to make extra payments … … leading to a significant difference in mean relative last billing rate % of customers with refunds and extra payment in last billing rate Mean last billing rate in % of total yearly rate % of customers (difference significant at 1% level) 63 56 47 Churners Tariff switchers +0. 7% Passive customers Extra paymen t made -2. 3% 37 44 Churners Tariff switchers 53 Refund receive d Passive customers -5. 1% Reading example: People who churned are those that had to make an extra payment of 5% of their total yearly billing rate to complete their last billing cycle. Note: all differences to passive customer sample significant at 1% Analysis excluding outliers with last billing rate >100% or <-100% of total amount 19

Results hold if we control for other variables, but high refunds can also have

Results hold if we control for other variables, but high refunds can also have negative effects Results of multinomial logit model: three variables (stay passive (basis), churn, switch) Payment sequence information Customer information Model fit Model 1: Refund sequence dummy model Odds-ratios: Churn Tariff switch Refund sequence dummy Model 2: Asymmetric magnitude model 0. 627*** 0. 788*** - - Magnitude of refund sequence - - 1. 530* 1. 708*** Magnitude of extra payment sequence - - 8. 489*** 4. 187*** Length of customer relationship (in month) 0. 878*** 0. 883*** 0. 879*** 0. 883*** Average price per k. Wh paid (in €) 0. 615*** 1. 010 0. 576*** 0. 995 Total usage (in k. Wh/yr. ) 1. 076*** 1. 060*** 1. 075*** 1. 060*** Nagelkerke's R-Square 0. 326 0. 328 -2 Loglikelihood 28, 025 28, 076 Chi-Square 6, 731 6, 777 N 22, 921 ***p<0. 01; **p<0. 05; *p<0. 1 H 4 a/b: Refund sequences have a negative influence on on churn / tariff switching probability High refunds may also have negative effects, but effect of high extra payment 4 x as large 20

Findings & implications Question 1: Do preferences for payment sequences exist? Question § Findings

Findings & implications Question 1: Do preferences for payment sequences exist? Question § Findings § Preference for refund sequences exist and people are willing to pay more just to experience refund. Preference for refund sequences decreases with the relative size of the refund to the extra payment Question 2: Question 3: Do payment sequences have attitudinal consequences? Do payment sequences have behavioral consequences? Refund sequences … § … decrease price consciousness § … decrease churn and tariff switching probability … § … (increase likelihood to recommend provider) § … but refunds should not be too high 21

Implications of our research Managerial implications Theoretical implications § Advance payment systems could be

Implications of our research Managerial implications Theoretical implications § Advance payment systems could be a viable alternative to post-payment or pre-payment systems § Advance payment systems may be subject to further research to identify preferences between different payment systems § Advance payment rates should be set such that chance of receiving a refund is increased § § However, there is a limit to how much providers should overcharge Commonly accepted finding that customers show more "rational" payment timing preferences for utilitarian goods (Prelec & Loewenstein, 1998; Patrick & Park, 2006) does not hold in advance payment systems 22

Managerial Implication: Purposely aim for refunds! Advance payments are adjusted every year…. … mostly

Managerial Implication: Purposely aim for refunds! Advance payments are adjusted every year…. … mostly leading to a zero-centered distribution of last billing rate Single customer example Last billing rate in 2011 -€ 190 Change of monthly advance payment for 2012* -€ 18 Last billing rate in % of total amount due** Increase in advance payment rate 5% (i. e. , 3. 45€ per month on average) Share of customers receiving a refund 70% Decrease in churn 7. 7% * Note: Change = Refund / 11 such that 11*monthly advance payment equals total billing rate for 1 year ** Source: Customer sample from survey 2 (cut-off at +/- 100%): N=840 23