Does the revenue increase if more payment methods are accepted?
Is there an influence of accepting cash and card payments instead of onlycard payments on the revenue of a restaurant?
Bachelor thesis
Economics and Business Finance and Organization Student: Daan Appel Student number: 10286047
2 Statement of Originality
This statement is written by Student Daan Appel who declares to take full responsibility for the contents of this document.
I declare that the text and the work presented in this document is original and that no sources other than those mentioned in the text and its references have been used in creating it.
The faculty of Economics and Business is responsibly solely for the supervision of completion of the work, not for the contents.
3
Table of contents Page number:
1) Introduction 4
2) Theory 6
3) Literature 7
4) Hypothesis 13
5) Data and Methodology 15
6) Results 20
7) Discussion 26
8) Conclusion 28
4 1. Introduction
At November the 5th 2015 the Volkskrant published an article: “Card payments almost overtake cash payments”. This article based its title on a research done by the Dutch national bank. This paper showed that the amount of card payments jumped from 40% in 2013 to 45% in 2014. The cash payments
decreased from 57% in 2013 to 53% in 2014. If this trend continues then 2015 or 2016 will be the first year that card payments overtake cash payments. One of the reasons for this increase is the
introduction of contactless card payments. For contactless payments a pin code is not required for payments under 25 euro, this could lead to an increase of card payments under 25 euro. Many customers nowadays are used to pay large amounts of money by card and small amounts in cash. The point where this preference changes is fifteen euros. (DNB, 2014)
This paper will discuss the influence of accepting cash and card payments or accepting only card payments on the revenue of a restaurant. In the current literature there are several references to the payment method and the way customers determine how to pay. To determine how customers pay the literature distinguishes several factors.
The most important factor that determines what the payment method will be is the transaction characteristic. Bounie and Francois (2006) show in their paper that almost 66% of the food and
beverages transactions are paid in cash. They also show the significant factor of transaction size. These results are also shown in research done by Klee (2008), the Dutch central bank (2014) and the European central bank (2014). Another factor that determines the payment method is the demographics. Carrow and Staten (1999) and Klee (2008) show the importance of two demographic factors, the education level and age. They show that the older the customer, the more likely it is that they will pay cash. Carrow and Staten (1999) show that the education level has a negative correlation with the usage of cash. A higher level of education indicates more card payments. Another factor which has an influence on the payment method is the type of good. Bounie and Francois (2006) show that the type of good has a significant effect on the payment method. They show that purchases related to food and beverages are more likely to be paid in cash than by card payments. This is also shown in a paper conducted by the Dutch national bank (2014). They show in their paper that 68 percent of the payments in hospitality are paid in cash. Another factor to determine the payment method choice is the cost of the payment method. This factor is not only important for the customer, but also for the merchant. Higher costs of the payment method lead to a lower probability of using this method. Schuh, Shy and Stavins (2010) show the differences in time per transaction. The time per transaction leads to differences in wage per transaction. Other costs for the merchant and the customer are cardholder fees. Some other factors that determine the payment
5 method and are covered in the literature are safety, budget control and technology adaption.
This paper measures the influence of payment methods on the revenue of a restaurant. In the current literature there does not exist a study that used a similar dataset in the hospitality sector. In Humphrey (2008) they ask for further research concerning the importance of cash as a payment method. Kahn and Roberds (2006) concluded that the empirical work is still in its infancy. This paper extends the current literature by providing a unique dataset. The real-life dataset will be provided by the bar-restaurant of The Student Hotel. This restaurant has a young, high educated audience. The dataset consists of the total revenue, the revenue per product group, the revenue paid in cash, the revenue paid by card and the cash/card payment ratio. To measure the influence of the acceptance of cash, the revenue will be compared before and after the acceptance of cash. To measure the influence of product groups, the difference in revenue per product group will be tested against the difference in total
revenue. There is research done about the influence of payment methods on the payment
characteristics, but these papers used data conceded from grocery stores and gas stations. In this paper the data is provided by the bar-restaurant of The Student Hotel. The dataset used in this paper is from a bar-restaurant and as shown by the Dutch National Bank (2014), the ratio of payment method differs significantly per sector. At gas stations in the Netherlands the proportion paid in cash is 29 percent and in the hospitality sector it is 68 percent. Another unique point of this paper is that the restaurant was cash free since the beginning, but started accepting cash several months ago. This unique and real-life dataset separates this research from the current literature and shows the influence of accepting cash.
6 2. Theory
This paper measures the influence of the acceptance of cash on the revenue.
To determine how the revenue could be influenced, first will be decided how customers choose their payment method. Second, will be determined how this decision can influence the revenue.
The current literature shows that there are many factors that influence the customer’s choice to determine which payment method to use. The three most important factors are: transaction size, demographics and type of good. Bounie and Francois (2006) show that small transactions are more paid in cash than by card. Carrow and Staten (1999) show in their paper that consumers are more likely to use cash when they have less education, are middle aged and own fewer credit cards and Bounie and Francois (2006) show that the type of good also has a big influence on the decision of which payment method to use. The transaction size and the type of good are positively related to the use of more cash in a bar-restaurant. As mentioned in the introduction, research done by the Dutch central bank shows that 68 percent of the transactions in the hospitality sector is paid in cash and that the average payment in cash is twelve euro twenty. These findings suggest that, once accepted, the cash payments will be a significant amount of the total payments. If a restaurant does not allow customers to pay cash this could lead to a situation where customers can not buy products, because they do not have a card with them. After the acceptance of cash, customers can choose which payment method they prefer and if they only have cash or only have a card with them they still can purchase products. When a restaurant offers cash payments this could lead to an increase of payments with a small transaction size, because most of the small sized transactions are preferably paid in cash. Because customers prefer to pay food and
beverages in cash, this might lead to an increase in sales of these types of goods. These two factors might lead to higher revenue. If a restaurant does not allow cash payments it could lead to customers who are not able to pay with their preferred method and this could lead to less satisfied customers. In the long term this could lead to avoidance of the restaurant and a decrease in the revenue.
7 3. The current literature
The current literature analyses different payment options from different perspectives. As mentioned in the introduction the way the revenue is influenced by payment methods in the hospitality sector has not previously been researched. Most literature is written about the way the payment method is influenced by different factors. The factors discussed in this paper are: transaction size, demographics, type of good, costs, safety, budget control and technology adaption.
In the next section, the current literature is summarized and it is sorted on relevancy to this paper. It starts with the transaction size which has a major influence on the payment method and ends with subjects that have a smaller influence on the customer’s payment decision like safety and
technology adaption.
Transaction size
In the literature there are many articles about the way that transaction size influences the payment method. Bounie and Francois (2006) show that the transaction size has a significant influence on the payment instrument. The larger the transaction size is, the lower the probability of using cash. More precisely if the transaction size is 10% larger, than the probability of being paid in cash will be 7% lower. Similar results were found in Klee (2008). They calculated the mean value of sale for cash and card payments. Cash transactions are very concentrated at a low value of sale, with a median value of sale at about $14.20. For credit cards, the median value of sale is $30.85. In the paper ‘Betalen aan de kassa’ conducted by the Dutch Central Bank (2014) they show that the average payment in cash is €12.20 and the average debit card payment is €30.40. In June 2014 the European Central Bank published a paper about the cash usage of consumers. This showed a significant difference in the way customers paid based on the transaction value. In their paper they divided the transactions based on their value in four quartiles and showed the differences in payment methods (cash, debit card and credit card). They showed these differences for seven different countries and it showed that only in the Netherlands the non-cash payments in the first quartile are higher than 10% of the total transactions (14%). For all countries, they found that the cash payment share is higher than 50% up to the median transaction value. In the third and fourth quartile, the dominance of cash fades. In the third quartile, however, cash has a higher share than debit or credit in three countries and a share that is about equal to the share of debit for CA, FR, NL, and US. In the fourth quartile in six from the seven countries they showed that card payments has a higher share than card payments.
8 Demographics
Another important factor how the payment methods are influenced is the demographics. Carrow and Staten (1999) show in their paper that consumers are more likely to use cash when they have less education, are middle aged and own fewer credit cards. These results are also shown by Klee (2008). In that paper the results show that age groups over 35 year are significantly more likely to use cash relative to the age group under 35. Klee (2008) also shows that education is an important variable to determine whether the payment will be paid in cash or otherwise.
Turning to the education results, higher education implies that the consumer is on average less likely to use cash or check and more likely to use credit cards. The European Central Bank paper (2014) shows that the results concerning the age hold in all the countries, except for the United States. This paper also shows that in all the countries the cash payments are lower when the income is higher. When consumers are higher educated they spent less money cash and pay more by card.
Type of good
Another factor that determines the payment choice is the type of good. Examples of different type of goods are gasoline, cold drinks, snacks or clothes.
Bounie and Francois (2006) show that the type of good has a significant effect on the use of a payment instrument but this effect varies across payment instruments. Globally, they found that the probability of using cash for purchases related to “Food and beverages” is higher than that of bank cards or checks. This is consistent with the results that the Dutch national bank found in their paper ‘Betalingsverkeer Kassa’ (2014). They show that in the hospitality sector over two third of the transactions is paid in cash.
Costs
The fourth important factor to determine the payment method is the cost difference with the different payment method. The higher the costs of a certain payment method the lower the probability will be that the customer chooses for this method. Faced with many choices: cash, check, debit or credit card, etc. consumers naturally consider the costs and benefits of each payment instrument and choose accordingly. For credit cards, the most important benefit is that consumers have delayed payments. They can buy an article on credit. Schuh, Shy and Stavins (2010) show that the costs for credit card payments are reflected in the price for customers. Humphrey, Kim and Vale (2001) show the benefits of card payments in contrast to cash payments for a country’s total economy. Especially in the long run
9 card payments will be beneficial for the economy. Humphrey, Pulley and Vesala (2000) estimate the cost of payments in U.S. to be as high as 3 percent of GDP. Since an electronic payment often only costs from one-third to half as much as a paper-based transaction considerable social benefits can be realized by promoting the use of electronics. Ardizzi (2013) shows that payments by card also have disadvantages like a cardholder fee and that the cardholder fee for the merchant sometimes is calculated into the price that customers have to pay.
Klee (2008) shows the importance of transaction time. The longer a transaction takes the more salary costs the merchant has to make. The paper shows the difference in time per payment method. A cash transaction where the customer chose to tender exact change has an average transaction time of about 8.5 seconds transaction time. The results indicate some asymmetry between tendering coins and receiving coins as change. While tendering each coin adds considerable time to the transaction,
receiving coins as change saves a little time. There is a difference in transaction time between cash and card payments. Card payments consist of a subset of payment instrument handling costs are likely authorization and verification costs. Theoretical work by Whitesell (1989) suggests that authorization and verification costs of payment instruments, which are fixed costs of using these payment
instruments, can motivate merchants to use substitutes. If a transaction takes more time a merchant has to make more wage costs. Credit card transactions require signatures. Debit card transactions require a PIN, and cash transactions require no authorization. Verification costs for card payments usually involve a “ping” on a database of collected profiles of payment records. The time to pay is 15 seconds with credit cards, and 9 seconds with PIN debit cards. These calculations suggest that the time differential between signing and entering a PIN is about 6 seconds. Nowadays this difference can decrease because of the wireless debit cards. These cards do not require a PIN for payments under 25 euros. This makes it the fastest payment method under 25 euros.
Safety, budget control and technology adaption
Some other topics in the literature which have influence on the payment method are: safety, budget control and technology adaption. Safety is not only influential for customers to determine how to pay, but also for the merchant. Kosse (2010) describes the differences in preferences about paying in cash or by card in different settings. The results show that debit card usage is less preferred when consumers feel unsafe. Hayashi and Klee (2002) wrote about the way customers adapt to new
technology and new payment methods. The results indicate that consumers who use new technology or computers are more likely to use electronic forms of payment and that payment choice depends on the
10 characteristics of the transaction such as the transaction value and the physical characteristics of the points of sale (such as the absence of a cashier or the availability of self-service). Hernandez, Kosse and Jonker, (2014) show that, on average, consumers responsible for the financial decision making within a household find the debit card more useful for monitoring their household finances than cash. Individuals differ in major respects, however. In particular, low earners and the liquidity-constrained prefer cash as a monitoring and budgeting tool. They also write that cash and payment cards differ considerably in the way they allow the user to set pre-defined budgets and to monitor the amount left to spend. With cash one may stick to a predetermined budget by withdrawing the amount of money allowed to be spent during a particular period and paying only in cash.
Related studies
In the next section three different related studies will be summarized. These studies all show the influence of the payment method. Two papers use customer surveys for their datasets and one paper uses checkout transactions. In all the three papers they use a logit model. Two of the papers show that as the value of the sale increases, consumers make more payments by card. All the papers show that the use of cash is strongly correlated with the demographics of the customer. The older age groups are significantly more likely to use cash than the younger customers. Turning to the education results, higher education implies that the consumer is on average less likely to use cash or check, and more likely to use credit cards.
How people pay: Evidence from grocery store data by Klee (2008)
Klee (2008) showed empirical evidence from grocery store transaction data. The results show that as the value of the sale increases, consumers are more likely to use credit cards and less likely to use cash and debit cards. The probability of using credit cards decreases as income increases; however, the probability of using debit cards increases. Survey data indicates that income, age and demographics are significantly correlated with payment use. The older age groups are significantly more likely to use cash or check relative to the baseline head of household age, under 35. Turning to the education results, higher education implies that the consumer is on average less likely to use cash or check, and more likely to use credit cards.
The observations that are used in this paper are the checkout transactions. The checkout transaction represent one customer’s total purchase at the point of sale. The data comprise over 10
11 million checkout transactions over a three month period, from September to November 2001. At the grocery stores in the sample, consumers can pay with six different payment types, namely cash, check, credit card, debit card, WIC, and food stamps. The first four are used generally, but the last two are associated with government food programs; thus the analysis focuses on the first four only.
Klee (2008) used three different models in her paper. First, she estimated the transaction time for the different transaction methods. Second, she estimated the probability of choosing a payment method. The variables were the customer characteristics, transaction characteristics and an error variable. Third, the costs of the transaction, the payment instrument choice, and expenditure were linked by using a selection framework introduced by Dubin and McFadden (1984).
Consumer cash usage: A cross country comparison with payment diary survey data by Bagnall, Bounie, Huynh, Kosse, Schmidt, Schuh and Stix (2014)
In this paper the usage of cash is measured in seven different countries. The results show that in all seven countries considered, cash is still used extensively – particularly for low-value transactions. This paper demonstrates that the use of cash is strongly correlated with demographics and point-of-sale characteristics such as merchant card acceptance and venue. These results hold apart from transaction sizes and consumer preferences for ease of use.
They collected data by using payment diaries of customers. Each diary attempts to record non-business-related personal expenditures of the respondent. All respondents were asked to record: The date, the transaction value, the payment instrument used, and the merchant’s sector where the purchase occurred. The respondents were asked to assess whether the purchase could have been paid using payment instruments other than the one actually used.All diary surveys yielded data sets containing more than 10,000 transactions. The sample contains all individuals (also those without payment cards) and all transactions that are conducted using cash, debit, or credit.
The authors used a logit model to estimate the probability of choosing cash versus non-cash alternatives (either debit or credit) at the point of sale. The variables used in the regression are: transaction size, cash balances, socio- demographic characteristics (age, income, and education), consumer perceptions of ease of use, acceptance, and cost, and point of sale transaction characteristics (card acceptance and type of purchase).The goal of these estimations is twofold. First, the authors want to quantify which factors exert an impact on consumers’ choice of whether or not to pay in cash, even when controlling for other potential factors. Second, the authors would like to study whether the use of cross-country data reveals patterns that are common to all countries.
12 Debit, Credit, or cash: Survey evidence on Gasoline purchases by Carow and Staten (1999)
This paper analysed the consumer’s payment options. The results showed that consumers are more likely to use cash when they have less education, lower incomes, are middle-aged, and own fewer credit cards. Debit and credit card users are younger, more educated, and hold more credit cards. The data that they used was conducted by a mail survey of gasoline credit cardholders during the spring and summer of 1992. The Credit Research Center at Purdue University provided
questionnaires to twelve participating oil companies with proprietary credit card programs. The companies mailed 24,000 questionnaires to samples of their cardholder base. The particular company sending the questionnaire was not identified to the consumer. Consequently, the consumers’ responses relate to their general use of gasoline credit cards and not to their behaviour with respect to a specific company (unless they owned only one gasoline card). The overall response rate was 25.9% (6451 total surveys). Due to incomplete demographic and credit information, some surveys had to be excluded. In total 5164 useable responses were collected.
The authors used the generalized extreme value (GEV) model, also known as the nested
multinomial logit model. To analyse the respondent’s choice of cash, general purpose card, and gasoline card they used the GEV model. The use of the GEV model is a sequence of logit models. First, they analysed the estimates for a model of choice within a given subset (gasoline or general purpose credit card). Second, they used the sum of the utilities for all the items in the subset (also called the inclusive value) as an explanatory variable in a higher-level model of choice (cash versus any credit card). Independent variables in the model include demographic characteristics and credit characteristics.
13 4. Hypothesis
The literature clearly shows that transaction size, type of goods and demographics are significant to the decision of customers how to pay. This paper uses a dataset provided by the bar-restaurant located in The Student Hotel. Because of the characteristics of a bar-restaurant, a low transaction size and food and beverages as main type of good, the transaction size factor and the type of goods factor could lead to cash payments. The transaction size of the average payment in this restaurant is assumed to be low. The data will provide more information about this assumption. The Dutch national bank (2014) and Klee (2008) show that the average cash payment is twelve euro twenty and show that the average card payment is thirty euro forty. These two results show that small size transactions could lead to cash payments and might lead to an increase of revenue because of the availability of their preferred payment method. Bounie and Francois (2006) show that the types of goods that are sold in a restaurant are significantly more paid in cash. Another factor which could lead to more cash payments is that in this sector there are more cash payments than card payment. This is shown in a paper conducted by the Dutch national bank (2014). They show that there are only two other sectors that have a higher ratio of cash payments than the hospitality sector.
The demographics of the customers in this paper could lead to a slower increase of cash payments and more card payments. Most of the customers in this paper are young students. As shown by Carow and Staten (1999) and Klee (2008) customers younger than 35 are more likely to pay by card. The education of the customers is also an important factor. Carow and Staten (1999) show that higher education leads to a preference of card payments.
Concluding, in this paper there are three important factors that could lead to cash payments and one important factor that could lead to a slower increase of cash payments. In this paper the hypothesis is that the acceptance of both payment methods leads to an increase of the revenue. This hypothesis is based on the multiple factors that could lead to cash payments. In this paper three factors could lead to cash payments and only one to a slower increase of cash payments, so there will probably be cash payments when this payment method is accepted.
The results from the literature lead to the following research question: is there an influence of accepting cash and card payments instead of only accepting card payments on the revenue of a restaurant. The first hypothesis is that the revenue after the acceptance of cash is larger than the revenue before cash.
14 This hypothesis will be tested for the total revenue and for all the product groups separately. Bounie and Francois (2006) showed that food and beverages are significantly more paid in cash. The acceptance of cash could lead to an increase in specific type of goods. In the data there are ten different product groups. The product groups are divided between six different drink groups and four food groups. The drink groups have a smaller average transaction size than the food groups. The expectation is that the change in revenue will especially be shown in the product groups with lower transaction value. These product groups are alcoholic drinks, warm drinks and cold drinks. The breakfast, lunch and dinner have higher average transaction values, so the expectation is that the food groups increase less than the drinks. To test the influence of the transaction size the increase in percentages of the drink groups will be compared to the percentages change of food groups. The second hypothesis is that the increase in revenue of drinks is larger than the increase in revenue of food:
𝐻1: 𝐼𝑛𝑐𝑟𝑒𝑎𝑠𝑒 𝑖𝑛 𝑟𝑒𝑣𝑒𝑛𝑢𝑒 𝑑𝑟𝑖𝑛𝑘𝑠 > 𝐼𝑛𝑐𝑟𝑒𝑎𝑠𝑒 𝑖𝑛 𝑟𝑒𝑣𝑒𝑛𝑢𝑒 𝑓𝑜𝑜𝑑
The alcoholic drinks, the warm drinks and the cold drinks are separated between students who live in The Student Hotel and customers who do not live in The Student Hotel. This separation shows a difference in demographics. The students have a lower average age and a higher education than the customers who do not live in The Student Hotel. Carow and Staten (1999) show that customers that have a lower age and a higher education prefer card payments over cash payments.
In the data there are three drink groups for only students and three drink groups for non-students. The three student groups are called: TSH hot drinks, TSH cold drinks and TSH alcoholic drinks. To measure the influence of demographics the students drinks are compared to the non-student drinks. The
expectation is that students prefer to pay by card compared to cash, so that the increase in revenue for student drinks will be lower than the increase in non-student drinks. In the dataset the student drinks are called TSH drinks and the non-student drinks are called drinks. The third hypothesis is that the increase in revenue of drink groups for non-students is larger than the increase in revenue of drink groups for students:
𝐻1: 𝐼𝑛𝑐𝑟𝑒𝑎𝑠𝑒 𝑖𝑛 𝑟𝑒𝑣𝑒𝑛𝑢𝑒 (𝐷𝑟𝑖𝑛𝑘𝑠) > 𝐼𝑛𝑐𝑟𝑒𝑎𝑠𝑒 𝑖𝑛 𝑟𝑒𝑣𝑒𝑛𝑢𝑒 (𝑇𝑆𝐻 𝐷𝑟𝑖𝑛𝑘𝑠)
15 5. Data and methodology
This part describes the data and methodology used. First the data used in this paper is described. Second, the method used to analyse the data is discussed.
The data is obtained from the manager food and beverages of The Student Hotel. The data contains the specific revenue numbers of all the locations that first only accepted card payments and now are accepting cash and card payments. This is a unique dataset as The Student Hotel moved from accepting only card payments to accepting cash- and card payments. This makes it possible to identify the effect of cash payment introduction on the revenue.
The data
The dataset is provided by The Student Hotel. Since the beginning in 2007 every bar-restaurant of The Student Hotel was cash free. Since a couple of months they started accepting cash payments. The exact date differs per location. The main reasons why they first did not accept cash was that it was easier for the employees and they thought that the revenue would not be affected. The reason to start accepting cash was that a lot of customers requested this change.
The Student Hotel
The Student Hotel offers fully furnished rooms for students and young professionals. Every student or young professional who lives in The Student Hotel is allowed to stay there for a time period of a minimum of three months up to ten months. All the locations of The Student Hotel offer hotel rooms for short stay guests next to the student housing. The ratio of hotel rooms to student housing is around two to five. In The Student Hotel there is no minimum age for renting student housing, but the average age is 21. The concept of The Student Hotel exists since 2007 when they opened their first location in Liège, Belgium. The Student Hotel currently operates over 2,750 rooms across six cities. This includes the two locations in Paris and Barcelona. They acquired these two locations after a merger with Melon District in 2015. The ambition of The Student Hotel is to operate over 10,000 rooms by the year 2020. The Student Hotel has two locations that are currently under development, Eindhoven and Groningen. The acquisition of a 20,000-m2 vacant office building in Florence was their last acquisition and was finalized in October 2015. Every location of The Student Hotel contains a bar-restaurant where students, hotel guest and locals can get food and drinks. The products include breakfast, lunch, dinner, drinks and snacks.
16 Different product groups
The dataset used in this paper contains the specific revenue numbers of different locations of The Student Hotel. The revenue is divided in ten different product groups and contains the revenue of four months before the acceptance of cash and four months after the acceptance of cash. Three groups are specifically for students who live in The Student Hotel. These groups are TSH alcoholic drinks, TSH hot drinks and TSH cold drinks and contain a ten percent discount on the drinks. These three groups have their equivalent for customers who do not live in The Student Hotel and do not have a discount included. This separation between students who live in The Student Hotel and non-students show the differences in demographics.
Except for the drinks there are four more groups: breakfast, lunch, dinner, snacks. The revenue is also divided per employee. To use the register, an employee has to log in using his or hers specific employee number. All the transactions that are made are accounted under his or hers account. This specification is not used in this paper.
Different Locations
The bar-restaurant of the location in The Hague is fully operational since mid-December 2014. The date that they started accepting cash was on June 17, 2015. The data from the location in The Hague includes four months before and four months after the acceptance of cash. The months before the acceptance of cash are March, April, May and June. The months after the acceptance of cash are September, October, November and December. July and August are excluded from the dataset, because in these two months the hotel has only hotel guests and no students. In the other months the ratio hotel guest compared to students does not significantly differs. The December month is limited to the 20th of December, because after this date a lot of students leave their apartment temporarily to visit their parents or go on Christmas holiday.
In Rotterdam they started accepting cash since the first of September. The bar-restaurant of the location Rotterdam is open since august 2014. The data for the location Rotterdam contains the same eight months of data as The Hague. The four months after accepting cash, September, October,
November and December and the months before accepting cash; March, April, May and June. The data of the location Rotterdam also contains the months September, October, November and December in 2014. These revenue numbers are also tested against the four months after the acceptance of cash.
17 These months are used because the dates and events are exactly the same, so there cannot be any differences based on events.
The data used of the location Amsterdam is different than the ones before. The location
Amsterdam is officially started accepting cash December the 20th 2015. The data after the acceptance of cash starts from January the 4th. January the third is the last day of the Christmas holiday and as shown before holidays are not used in this sample. For the location Amsterdam there is a comparison between the four months before the acceptance of cash and the four weeks after accepting cash. The difference in sample size is significant and will be marked in the results.
Outliers
The outliers in this dataset are based on the revenue. The days that have a difference of more than twice the average revenue per day are removed from the dataset. The significant difference in revenue had several reasons. One of the reasons the data consisted of outliers was the vending machine. In the vending machine The Student Hotel sells drinks, candy bars and other snacks. The revenue of the vending machines was added to the dataset once in a couple of months. The revenue of a vending machine is around €1500 a month, so this extra revenue could multiply the daily revenue. The revenue of the vending machine is reduced of the total revenue of that specific day, so the day is still used in the sample. Another reason that significantly increased the revenue was the events. An event in the hotel increases significantly the revenue of the bar-restaurant every time. Because the events were hosted every time in a different way, these days are removed from the dataset. The events around the bar-restaurant often had a significant influence on the revenue. Days that have an increase of twice the average revenue because of an event are removed from the dataset. The last reason for the outliers is the difference in opening hours. Days with different opening hours are removed from the sample size as well, because longer or shorter opening hours lead to unmeasurable differences. Most of the days had different opening hours based on events in or around the hotel. In total five outliers per location are removed.
18 Methodology
To analyse the data in this paper the revenue is compared before and after the acceptance of cash. The difference in revenue is measured in three different ways. For every location the total revenue is analysed before and after the acceptance of cash. This shows the total change of the revenue and if this change is significant. The test used to measure this significance is a t-test.
The revenue is also analysed per product group. The difference in revenue per product group can show more about the transaction size and the type of good. This difference is analysed twice with the t-test. In the first test, the difference will be analysed with the revenue numbers as shown in the data. In the second test, the revenue per product group will be tested against a corrected revenue number. This corrected revenue number is the revenue per product group before the acceptance of cash multiplied by the difference in revenue of the total revenue after accepting cash.
This new revenue number shows the revenue of the product group as if it changed the same percentage as the total revenue. If this test shows a difference, than this product group changed differently from the total revenue.
The influence of the demographics will be analysed as well. The data contains drink groups that are separated between students who live in The Student Hotel and customers who do not. A change in these revenues can indicate the significance of different demographics. The drink groups are divided in alcoholic, warm and cold drinks.
The paired t-test
To determine this difference the paired t-test will be used. This test is used because it measures the difference in averages for the paired samples. In this paper the average will be calculated of the different revenues before the acceptance of cash and after the acceptance of cash. The sample contains four months of data before and after the acceptance of. The sample size differs per location, but it is around 111 days. To measure the difference in revenue for different product groups the paired t-test will be used as well. The sample size before the acceptance of cash is four months. In the dataset it shows that during events in the hotel the revenue is significant higher than on days without events. To find this difference in revenue the event list was requested at the different locations and based on this information these outliers are removed from the sample. The sample also has different revenues on days with smaller or longer opening hours. Days with different opening hours are days with events in or
19 around the hotel and during national holidays. These days will be removed from the sample as well. In total five outliers per location are removed.
The results from the paired t-test will be presented and tested on their significance. To determine the effect the acceptance of cash payments had on the revenue, not only the revenue numbers will be presented but also the ratio of cash/total payments will be presented. This ratio shows how many transactions relative to the total transactions are paid in cash. An increase of this ratio shows the increase of cash payments. The results show the difference in revenue and ratio of cash/total payments for four months before the acceptance and four months after this acceptance.
20 6. Results
In this part the results will be presented. The results in this paper are all confidential and cannot be used for further research. First the difference in total average revenue will be presented. The results are presented in a graph and in specified tables. The tables will be separated between the different locations. Second, the difference in revenue per product group will be presented. The revenue per product group will also be corrected and tested for the change in total revenue. The difference in revenue per product group can show more about the transaction size and the type of good. Third the difference in revenue of student drinks and non-student drinks will be presented. This difference will be used to show the influence of demographics. The product groups contain customers with a different average and a different education level. The average revenue increase in this sector was eight percent on a yearly base.
Total revenue
First the difference in total revenue for Rotterdam and The Hague will be presented. The months used are March till June and September till December. The two locations are shown in a graph together and are specified in a table per location. The graph shows the total revenue per day before and after the acceptance of cash. On the vertical axis the revenue in euros is shown and on the horizontal axis the different days are shown. The tables show the average revenue per day based on all days in that month.
Before: March – June 2015 After: September – October 2015
0,00 500,00 1.000,00 1.500,00 2.000,00 1 8 15 22 29 36 43 50 57 64 71 78 85 92 99 10 6 11 3 12 0
Den Haag Rotterdam
0,00 500,00 1.000,00 1.500,00 2.000,00 1 8 15 22 29 36 43 50 57 64 71 78 85 92 99 10 6
21 Specified revenue
Rotterdam March 2015 – June 2015 Rotterdam September 2015-December 2015 Average revenue per day
March April May June €993.27 €1009.56 €1019.02 €1045.75 Cash/total ratio 0% 0% 0% 0%
Average revenue per day
September: October: November: December: €1337.00 €1402.50 €1378.99 €1280.72 Cash/total ratio 17% 33% 38% 43% Average €1020.15 Average €1385.94
The increase in average revenue per day before accepting cash and after accepting cash is: €365.79. The data contains 106 observations. The increase of the average revenue per day in percentages is 36.4 percent and is significant at 1%.
Den Haag March 2015 – June 2015 Den Haag September 2015 – December 2015 Average revenue per day
March: April: May: June: €343.03 €284.70 €326.92 €324.10 Cash/total ratio 0% 0% 0% 0%
Average revenue per day
September: October: November: December: €598.82 €505.07 €538.86 €549.43 Cash/total ratio 59% 55% 56% 53% Average €319.98 Average €547.09
The increase in revenue per day before accepting cash and after accepting cash is: €227.11. This dataset consist of 111 observations. The increase of the average revenue per day in percentages is 70 percent and is significant at 1%.
22 Next the revenue for the Rotterdam location will be tested again, but with different months now. The Rotterdam location will be tested with the same months as the year before. This shows the robustness and eliminates errors that could occur by comparing different months. The data contains 111
observations on both sides.
Rotterdam September – December 2014 Rotterdam September – December 2015 Average revenue per day
September: October: November: December: €1162.22 €1110.75 €1103.88 €1033.52 Cash/total ratio 0% 0% 0% 0%
Average revenue per day
September: October: November: December: €1337.00 €1402.50 €1378.99 €1280.72 Cash/total ratio 17% 33% 38% 43% Average €1069.97 Average €1385.94
The increase in average revenue per day before accepting cash and after accepting cash is: €315.97. The increase of the average revenue per day in percentages is 29.5 percent and is significant at 1%.
The location in Amsterdam will be tested with a different timeframe. The cash acceptance started at December 20th, the beginning of the Christmas holiday. The Christmas holiday is excluded from the dataset because of the different demographics of the customers during the holiday. The data contains four weeks of data before the acceptance of cash and four weeks of data after the Christmas holiday.
Amsterdam November-December 2015 Amsterdam January 2016 Average revenue per day
23/11/15 – 20/12/15: €1558.93
Cash/total ratio
0%
Average revenue per day
January 2016: €1747.01
Cash/total ratio
45%
The increase in average revenue per day before accepting cash and after accepting cash is: €188.08. The increase of the average revenue per day in percentages is 12 percent and is significant at 10%.
23 The change of revenue at location in Amsterdam is also tested for the four months before the
acceptance of cash. This dataset contains 114 observations in the months September till December.
Amsterdam September - December 2015 Amsterdam 2016 Average revenue per day
September: October: November: December: €2152 €1867.42 €1865.48 €1484.58 Cash/total ratio 0% 0% 0% 0%
Average revenue per day
January 2016: €1747.01
Cash/total ratio
45%
Average €1874.43 Average €1747.01
The decrease in revenue per day before accepting cash and after accepting cash is: €127.42. The decrease of the average revenue per day in percentages is 5.7 percent and is not significant for 10%.
The revenue separated by different product groups
In the next part the revenue is analysed by the different product groups. The dataset used to measure the difference in revenue from product groups is the data from the Rotterdam location. This dataset contains 111 observations and uses the months: October, September, November and December in 2014 and 2015. The difference is measured by using a t-test to compare the average before the acceptance of cash and the average after this acceptance. The revenue per product group will also be tested against a corrected revenue number. The corrected revenue number shows the revenue of the product groups increased the same percentage as the total revenue.
24 Student drinks and non-student drinks
First the drink groups are presented and second the other groups. The tables contain the different product groups, the revenue before the acceptance of cash, the corrected revenue and the revenue after cash. The asterisk after the revenue number after acceptance of cash shows if the increase in revenue is significant and/or significant for the corrected revenue. The TSH cold drinks is the only group that decreased in revenue, so it has a special symbol.
Product group: Alcoholic drinks Cold drinks Hot drinks TSH Alcoholic drinks TSH cold drinks TSH hot drinks Revenue before acceptance of cash: €63 €84 €133 €23 €47 €57 Corrected revenue: €81.6 €108.8 €198.1 €29.8 €60.9 €73.81
Revenue after acceptance of cash: €104** €111* €199* €35** €37ᵈ €99** * significant 5%
** significant 5% for corrected revenue ᵈ significant decrease
Dinner, lunch, breakfast and snacks Product group: Dinner Lunch Breakfast Snacks Revenue before acceptance of cash: €276 €199 €79 €63 Corrected revenue: €357.4 €257.7 €102.3 €81.6
Revenue after acceptance of cash: €337* €223* €77 €69* *significant 5%
25 Eight of the ten different groups show a significant increase in revenue after the acceptance of cash. Breakfast shows a decrease and TSH cold drinks shows a significant decrease. After the correction for the increase of the total revenue these results change. Three product groups increased significantly and one product group decreased significantly. The six other groups did not have a significant change.
Except for the TSH cold drinks, all the drink groups show significant increases. After the correction for the average revenue only alcoholic drinks, TSH alcoholic drinks and the TSH hot drinks show a significant increase. The comparison in the hypothesis is between the drink groups and the other groups. The other four groups all show a non-significant change compared to the corrected revenue. The importance of transaction value is shown by this comparison, because the drink groups have a low transaction value and the other groups have a higher transaction value.
The difference in revenue in the drink groups shows the influence of the demographics. In the literature it is shown that young people pay more transactions by card and adapt to new technologies easier. These results could lead to a slower increase of cash payments and a slower increase of the revenue, because young people pay more transactions by card than in cash. The significant increase of the two student drink groups does not confirm the third hypothesis that the student drink groups would increase less than the non-student groups. The three non-student drink groups increased, but corrected for the change in total revenue the results only the alcoholic drinks showed a significant increase.
26 7. Discussion
In this part the results and the methodology will be discussed. The hypothesis of this paper is how the acceptance of cash and card payments influences the revenue. First the literature is discussed. The current literature contains many subjects that influence the way customers pay. The literature shows different factors that determine the payment method. The most important factors are the transaction size, the type of good and the demographics. These factors are analysed in this paper. The data contains the total revenue, the revenue of different product groups, the revenue in cash, the revenue in card and the cash/total payment ratio. The product groups contain import characteristics. The drink groups have a significant lower transaction size than the breakfast, lunch or dinner group. The data shows that the biggest increase in revenue is found in the groups with the lowest transaction size, the drinks. To measure the influence of demographics the drink groups are used. The drink groups are divided into six different product groups. Drink groups for students who live in The Student Hotel and customers who do not live in The Student Hotel. Only students who live in The Student Hotel are allowed to buy TSH drinks, these drinks contain a ten percent discount. The students have a lower average age and a higher education than the hotel guests.
In the current literature there is no research done with a dataset comparable to the one used in this paper. The dataset in this paper uses real data about the revenue from a company that is not publicly traded. The dataset is confidential. The main reason that there is no dataset like this in the current literature is the confidentiality. To obtain a dataset like this is difficult. Most literature uses publicly available data. The advantage of a publicly available dataset is that it is possible for everybody to check and analyse the data.
Data
The data used in this paper is provided by the bar-restaurant of The Student Hotel. The data contains three different locations and the timeframe differs per location. It is two months for the location in Amsterdam and eight months for the locations in The Hague and Rotterdam. The months compared in Den Haag are the last four months before the summer holiday and the four months after the summer holiday. In Rotterdam the data contains the same months in The Hague and also the last four months of the year of 2014 and the last four months of the year of 2015. In Amsterdam it compares the four weeks before the Christmas holiday of 2015 with the four weeks after the Christmas holiday. The data contains the revenue per day, the revenue per product group, the revenue in cash payments, the revenue in card payments and the ratio of cash payments in relation to card payments.
27 Improvements
The data could improve by containing more locations. There are seven different locations of The Student Hotel right now. Of these seven locations only three locations can be used for this paper. Two locations accepted cash from the beginning, which made it incomparable. Two locations have a different bar-restaurant than the ones used in the sample, because they are only part of The Student Hotel concept since April 2015, when Melon District merged with The Student Hotel. The data can improve as well by containing a bigger timeframe. The data used in the location Amsterdam contains a month of data before the acceptance of cash and a month of data after the acceptance of cash, which is not a big timeframe for a comparison of average revenue. To get the best results out of the data, the data should be compared in every location, in the same months.
The data could further more improve by containing more variables. The variables in the current data set are all derivatives of the revenue in the different locations. One of the variables that could add more significance is the occupation rate of the hotel. If this is included, the revenue could be less influenced by different occupation rates.
Methodology
The methodology could improve by correct the revenue for more variables. In this paper the revenue is used based on the revenue numbers provided by The Student Hotel. The outliers of the revenue are deleted and the cash/card ratio is provided next to the results, but the revenue is not corrected for different variables. Variables used for this correction could be the participation rate in the hotel and the demographics of the guests. In this paper the absolute revenue differences are almost all significant. One of the reasons for this significance could be that there are unexplained variables included in the revenue numbers. More data and more variables could lead to less differences and a more balanced outcome. Another aspect that could have an influence on the revenue is the comparison of different months. It could be that different months have different average revenues, this could lead to biased result because this difference in revenue is not influenced by the payment method.
28 8. Conclusion
The literature shows that there are many factors that influence the payment method. The three most important factors are: transaction size, demographics and type of good. The literature shows that there is a small difference between countries, but no significant difference. The costs of the payment method are especially important for the merchant and not significantly important for the customer. The results in this paper show a significant increase in the total revenue of two of the three different locations. One location shows a not significant decrease of the total revenue. This confirms the first hypothesis that the revenue would increase after the acceptance of cash. The increase in revenue is separated in different product groups. Of the ten different product groups eight show a significant increase. Two product groups decrease in revenue, TSH cold drinks decreases significantly and breakfast decreases not significantly. These results show when the revenue is not corrected for the percentage of the total increase of the revenue. When the revenue before the acceptance of cash was multiplied by the percentage of the total increase, the results change. The results after this correction show that four of the ten groups have a significant difference with the revenue before the acceptance of cash. Three groups increase significantly and one group decreases significantly. The three product groups that are significantly increased are three drink groups. This increase of the revenue confirm the second
hypothesis, because these groups all have a small average transaction size and the food groups do not increase significantly. The difference between students and non-students shows in the hot, cold and alcoholic drinks. The hot and alcoholic drink groups for students increase significantly. The cold drinks for students decreases. These results do not confirm the hypothesis. The students have a lower average age and are highly educated compared to the other customers. The literature shows that these
demographics would probably lead to a slower increase of cash payments and a slower increase of the revenue. The bigger increase of student drinks than non-students drinks does not confirm these results.
Further research could show if this trend of increasing revenue continues. The dataset needs to be larger for further research. More similar companies or more months could be added to the dataset for better results.
This paper shows evidence of an increase in revenue after accepting cash payments. The increase is specified in different product groups. The product groups with the lowest transaction value show the biggest increase in revenue. The product groups with the highest transaction value show a smaller increase in revenue than the increase of the total revenue. This confirms the existing literature and the hypotheses.
29 9. References
Bagnall, J., Bounie, D., Huynh, K. P., Kosse, A., Schmidt, T., Schuh, S & Stix, H. (2014). Consumer Cash Usage A Cross-Country Comparison with Payment Diary Survey Data, , Working Paper Series, 1685.
Bounie, D. and François, A. (2006). Cash, Check or Bank Card? The Effects of Transaction Characteristics on the Use of Payment Instruments, Working Papers in Economics and Social Science.
Carow, K. A. and Staten, M. E. (1999) Debit, Credit, or Cash: Survey Evidence on Gasoline Purchases,
Journal of Economics and Business, 51, pp. 409–421.
Hancock, D. A. and Humphrey D. B. (1998). Payment transactions, instruments, and systems: A survey,
Journal of Banking & Finance 21, pp. 1573-1624.
Hayashi, F. and Klee, E. (2003). Technology Adoption and Consumer Payments: Evidence from Survey Data, Review of Network Economics Vol.2, Issue 2.
Hernandez, L. and Zwaan, P. (2014). Betalen aan de Kassa, De Nederlandse bank.
Humphrey, D. B., Pulley, L. B. and Vesala, J. M. (1996). Cash, Paper, and Electronic Payments: A Cross-Country Analysis, Journal of Money, Credit and Banking, Vol. 28, No. 4, Part 2: Payment Systems
Research and Public Policy Risk, Efficiency, and Innovation, pp. 914-939.
Humphrey, D.B. (2004). Replacement of cash by cards in U.S. consumer payments. Journal of Economics
and Business, 56, pp. 211–222.
Humphrey, D. B. (2010). Retail payments: New contributions, empirical results, and unanswered questions, Journal of Banking & Finance 34, pp. 1729–1737.
Kahn, C.M. and Roberds, W. (2009). Why pay? An introduction to payment economics, Journal of
30 Klee, E. (2008) How people pay: Evidence from grocery store data, Journal of Monetary Economics, 55, pp. 526–541.
Kosse, A. (2010). The safety of cash and debit cards: a study on the perception and behaviour of Dutch consumers, DNB Working Paper, 245.
Rochet, J.C. and Tirole, J. (2002) Cooperation among Competitors: Some Economics of Payment Card Associations, The RAND Journal of Economics, pp.549.
Schuh, S., Shy, O. and Stavins, J. (2010). Who gains and Who loses from Credit Card Payments? Theory and Calibrations, Public Policy Discussion Papers, 10.
