Time to Eliminate the Penny from the U.S. Coinage System: New Evidence

By Robert Whaples

Annotated on Fermat's Library with 1 comment. Text and annotations as Markdown Original paper

More detailed data about the number of transactions in a typical ba...
139
Eastern Economic Journal, Vol. 33, No. 1, Winter 2007
Robert Whaples: Department of Economics, Wake Forest University, Winston-Salem, NC 27109. E-mail:
whaples@wfu.edu.
TIME TO ELIMINATE THE PENNY
FROM THE U.S. COINAGE SYSTEM:
NEW EVIDENCE
Robert Whaples
Wake Forest University
INTRODUCTION
In “Eliminating the Penny from the U.S. Coinage System: An Economic Analysis,”
Raymond Lombra [2001] notes several adverse effects of eliminating the penny. He
argues that elimination of the penny will a) impose a sizeable and regressive “rounding
tax” on consumers, b) boost the infl ation rate and federal government outlays, c) drive
up fi rms’ costs and squeeze their profi ts, and d) deprive the government of consider-
able seigniorage. In this note I reexamine these fi ndings in light of new evidence and
conclude that it is probably time for the U.S. Mint to cease minting pennies.
NEW FINDINGS ON THE “ROUNDING TAX”
The heart of Lombra’s analysis is his estimate of the losses to consumers if the
penny were eliminated and cash prices were rounded up or down to the nearest
nickel. Using data from the price book of a typical convenience store, he shows that
the last digit for over eighty percent of prices is a nine. Simulating 5000 purchases of
from one to three items across the full spectrum of the 3585 items in the price book,
he estimates that 60 percent of the transactions result in rounding upward if three
items are purchased and up to 93 percent are rounded upward if one or two items
are bought. As a lower bound, if half of the 106 billion annual consumer transactions
were made with cash and 60 percent were rounded upward, consumers would col-
lectively lose $318 million per year. As an upper bound, if 83 percent of transactions
use cash and 93 percent have prices rounded upward, the loss to consumers would
be $818 million annually.
The fi rst limitation of this analysis is that it is unknown how many items (and
in what combinations) consumers purchase from convenience stores. The “rounding
tax” is lower as more items are purchased, but information on how many items are
purchased was unavailable to Lombra and many purchases involve more than three
items. More problematically this calculation assumes that all such transactions are
not taxed. Chande and Fisher’s [2003] simulations show that the “rounding tax” es-
sentially disappears in Canada where all items are taxed. Convenience stores’ goods
are taxed in the United States, too. According to the Federation of Tax Administra-
140
EASTERN ECONOMIC JOURNAL
tors [2006], only fi ve U.S. states (Alaska, Delaware, Montana, New Hampshire, and
Oregon), with 2.5 percent of the U.S. population, have no sales tax. On the other hand,
many items purchased at convenience stores are food items which are exempt from
sales tax in most states. States exempting food from sales tax made up 71.7 percent
of the U.S. population in 2006.
1
Given these complexities, it is unclear how rounding current prices would affect
American consumers. Fortunately, I was able to obtain data on nearly 200,000 indi-
vidual transactions from a convenience store chain that operates in seven states in
the eastern U.S. – Alabama, Georgia, North Carolina, Pennsylvania, South Carolina,
Tennessee, and Virginia.
2
The general sales tax and food sales tax rates for these
states are given in Table 1.
TABLE 1
Sales Tax Rates in Seven States (July 2005)
State Sales Tax Rate Sales Tax on Food
Alabama 4 4
Georgia 4 0, but subject to local sales tax
North Carolina 4.5 0, but subject to local sales tax
Pennsylvania 6 0
South Carolina 5 5
Tennessee 7 6
Virginia 5 3.5
Table 2 reports the gains and losses during a one-week period to this chain’s
consumers if their bills were to be rounded to the nearest nickel. The fi rst column
reports all transactions, including both cash and non-cash transactions. The second
column reports all transactions on which sales tax was charged – thus excluding
gasoline-only (and some other) purchases. The data set mixes together cash and non-
cash transactions, but there are a couple of close proxies for cash transactions in the
third and fourth columns of Table 2. The best proxy is probably transactions in which
change was returned.
3
Another proxy is transactions in which the total amount paid
is less than $5.
As the table demonstrates, the total effect on buyers and sellers from rounding
current totals to the nearest nickel would be very, very small. Across twenty stores
and 185,714 transactions, consumers win – but the total gain to customers is only
$105.60 or about one-twentieth of a cent per transaction. This amounts to a tiny loss
of 75.4 cents per store per day. In every state the results are similar. In a few states
consumers lose a hair, in the rest they gain a hair or two. The largest margin is a
customer gain of about one-eighth of a cent per sale in North Carolina.
Pennsylvania may be the most nationally-representative of these states, since it
does not charge sales tax on food. In Pennsylvania, customers do lose from rounding,
but the losses are not more than half a cent per transaction as Lombra’s simulation
implies. Consumers’ losses in Pennsylvania add up to an infi nitesimal $1.79 or one
cent for every 67 transactions.
Lombra’s analysis focuses on cash transactions, but these don’t appear to differ
from overall transactions. For the transactions in which change was returned to
141
TIME TO ELIMINATE THE PENNY
the customer – almost all of which would have been cash transactions
4
– customers
collectively would have gained $23.82 from rounding to the nearest nickel or about
one-fortieth of a penny per transaction. Likewise, customers would win marginally
from rounding of purchases that come out to less than $5, with a collective gain of
$28.87 or 1/26
th
of a cent per transaction. The bottom line is that rounding would
have essentially no effect on consumers or sellers as a group, if these transactions
are indicative. The “rounding tax” is effectively zero.
TABLE 2
Customers’ Gains and Losses from Rounding to the Nearest Nickel
State All Transactions: Transactions with Sales Tax
(# of stores) Column 1 Column 2
Gain/Loss to Total Gain/Loss Gain/Loss to Total Gain/Loss
Customers Transactions per Customers Transactions per
Transaction Transaction
Alabama (1) +$6.12 10,730 +0.057 cents +$0.51 5540 +0.009 cents
Georgia (4) +$30.18 43,885 +0.069 cents +$13.37 25,762 +0.052 cents
North Carolina (5) +$44.24 33,870 +0.131 cents +$34.69 22,681 +0.153 cents
Pennsylvania (1) -$1.79 12,034 -0.015 cents +$8.91 6080 +0.147 cents
South Carolina (4) +$19.98 41,275 +0.048 cents -$4.52 24,355 -0.019 cents
Tennessee (1) +$3.16 10,039 +0.031 cents +$1.06 6501 +0.016 cents
Virginia (4) +$3.71 33,881 +0.011 cents -$11.89 17,198 -0.069 cents
All Locations (20) +$105.60 185,714 +0.057 cents +$42.13 108,117 +0.039 cents
Akron Drive +$6.29 7001 +0.090 cents +$4.93 5255 +0.094 cents
Stratford Road +7.46 6591 +0.113 cents +$5.40 4223 +0.128 cents
State Change-Returned Transactions Transactions Less Than $5
(# of stores) Column 3 Column 4
Gain/Loss to Total Gain/Loss Gain/Loss to Total Gain/Loss
Customers Transactions per Customers Transactions per
Transaction Transaction
Alabama (1) +$1.44 5550 +0.026 cents +$0.47 3437 +0.014 cents
Georgia (4) +$9.13 23,576 +0.039 cents +$11.62 18,530 +0.063 cents
North Carolina (5) +$19.29 18,641 +0.103 cents +$26.53 15,775 +0.168 cents
Pennsylvania (1) -$4.84 7245 -0.067 cents -$2.88 5282 -0.055 cents
South Carolina (4) +$1.09 23,304 +0.005 cents -$5.57 17,064 -0.033 cents
Tennessee (1) +$2.06 6177 +0.033 cents +$3.06 4383 +0.070 cents
Virginia (4) -$4.35 16,029 -0.027 cents -$4.36 11,864 -0.037 cents
All Locations (20) +$23.82 100,552 +0.024 cents +$28.87 76,355 +0.038 cents
Akron Drive +$1.93 4626 +0.042 cents +$3.78 3608 +0.105 cents
Stratford Road +$3.88 3775 +0.103 cents +$3.48 2807 +0.124 cents
Again, in both of these cases customers in Pennsylvania would have lost a few
dollars – $4.84 on change-returned transactions and $2.88 on purchases of less than
$5 – but again the losses are very small – 0.067 and 0.055 cents per transaction, re-
spectively. Slicing the data more thinly reveals exactly which Pennsylvania custom-
ers would lose. Losses on purchases of less than $5 total $2.88, but losses on the 571
purchases of less than $1 make up more than all of this amount – $4.91. Customers
buying a single untaxed item whose price is less than $1.00 and whose price ends in
the 9 digit are the “big” losers. Those spending less than one dollar have a loss rate of
0.86 cents per transaction, but their collective loss is less than $5 per week at a store
142
EASTERN ECONOMIC JOURNAL
processing over 12,000 transactions per week. These customers are fairly atypical,
making up 4.7 percent of purchases.
Lombra’s conclusion that the tax would be regressive is also not supported by these
data. The bottom rows of Table 2 compare two stores in Winston-Salem, North Carolina.
The store on Stratford Road is in a fairly affl uent section of the city, while the store on
Akron Drive is in the poorer section of town. The rounding estimates are essentially the
same in both settings – with customers coming out slightly ahead in both locations.
5
IMPACT ON INFLATION
Lombra’s second major contention is that rounding would cause the infl ation
rate to rise slightly and boost federal government outlays by $1 billion per year or
more. This conclusion is mistaken because, as shown above, there appears to be es-
sentially no “rounding tax.” Even if there were a “rounding tax,” however, the calcu-
lation would be incorrect because eliminating the penny would at most (according to
Lombra’s calculations) introduce a one-time increase in prices of about 1/100 of one
percent. Such a small increase wouldn’t generally affect the Consumer Price Index
(CPI), which isn’t calculated this precisely. The Bureau of Labor Statistics reports
the CPI with only one decimal point. For example, in January, 2006 the CPI for all
urban consumers was 198.3 (1982-84=100). A 1/100
th
of one percent increase would
boost it to 198.31983, but the BLS would round this to 198.3, so there would be no
change. A 1/100
th
of one percent increase would have only about a one-in-fi ve chance
of boosting the index to the next signifi cant digit. Several federal expenditures and
revenues streams, such as Social Security and the income tax code, are indexed to
the CPI, but they calculate the infl ation rate to the tenths digit, so a 1/100
th
of one
percent increase would have only a one-in-ten chance of changing the offi cial infl ation
rate in one year, under Lombra’s scenario.
More importantly, the elimination of the penny would not stop prices from being
quoted in cents and non-cash payments – which are now more common than cash pay-
ments [Dove Consulting, 2005] – could and would still be made in cents, so the prices
recorded by the BLS wouldn’t change at all, nor would its calculation of the CPI.
6
IMPACT ON PROFITS AND COSTS
Lombra also concludes that rounding could hurt fi rms by squeezing their thin
profi t margins as it undermines the penny’s role in theft deterrence, boosts the costs
of retraining workers and spurs costly non-cash transactions. In response, sellers will
react strategically and raise prices.
Chande and Fisher [2003, 517] acknowledge that retailers could act strategically,
if they knew how frequently different combinations of items are purchased, but con-
clude that the amount gained per transaction “would be so trivially small as to have
little impact” on consumer behavior and welfare. Just as importantly, the convenience
store market is very competitive, so any change that affects all the fi rms in the market
– such as eliminating the penny – would have virtually no effect on their profi t levels.
Profi ts will tend to the normal level both with and without pennies. In competitive
industries prices refl ect costs, so prices would rise only if costs rise.
143
TIME TO ELIMINATE THE PENNY
In addition, it is unlikely that pennies and fi nal-digit nines have much of a theft-
deterrent role currently – because of taxes, few in-store transactions ever come out to
an exact dollar amount.
7
Likewise, today’s technology thwarts the kind of employee
theft that pennies may have once deterred – theft via the employee failing to register
the sale and pocketing the cash, which may be thwarted by the need to make change.
In most convenience stores such a ploy would be fairly obvious since customers are
conditioned to seeing the sales amount displayed on the register and hearing a “blip”
when each item is scanned. They often look to see if the register rang up the correct
amount, which is rarely marked on the package anymore.
The costs of retraining employees to operate in a penny-less economy, if they exist
at all, are likely to be minor. The employee need only follow standard procedure, scan-
ning items and waiting for the register to calculate how much is owed and how much
change is due. Countering any temporary learning costs, the elimination of pennies
would save a modicum of time spent fumbling with change – by both the customer
and the clerk – increasing effi ciency. Across the twenty stores in this sample, 36.3
percent of the transactions involved returning pennies in change. If the use of a penny
adds even one second to the transaction time, this represents 67,480 seconds or 18.74
hours of time for the clerk, the customer, any companions of the customer and anyone
in line. Average wages of retail clerks in 2004 were $9.39 per hour or 0.26 cents per
second [Bureau of Labor Statistics, 2005]. Using this wage, the value of lost time from
dealing with pennies among the clerks at these twenty stores totaled $175.97, which
dwarfs the estimates of the gains or losses from rounding. The average overall wage
(2004) was $18.09 per hour or 0.5025 cents per second, so the cost of the customers’
time wasted was $339 in this sample. Applying these rates to Lombra’s estimate of
106 billion transactions nationally each year and using the above estimates that 36.3
percent of transactions involve returning pennies and that returning a penny requires
about one extra second, this implies that the value of wasted time is 106 billion sec-
onds x 0.363 x (0.26 + 0.5025 cents per second) = $293.4 million. This sum is fairly
small – equaling about $1 per person per year – but it understates the cost somewhat
because it omits the impact on companions and others in line who have to wait and
it omits cases where pennies are tendered by the customer. If use of pennies adds 2.5
seconds to the average transaction, these losses reach $733.5 million per year.
8
Whether or not the elimination of pennies leads to a greater number of non-cash
transactions is an open question, but the marginal cost of such transactions has be-
come low enough that more and more retailers have voluntarily adopted credit and
debit payment for low-value transactions – even in fast food restaurants and vending
machines.
One reason given for the ubiquity of prices ending in nines is that they fool some
customers into thinking the item is substantially cheaper than it really is. Apparently
the gap between 99 cents and one dollar just seems larger than the gap between $1.00
and $1.01 to some customers. With the elimination of the penny, the perception that
99 cents is substantially less than one dollar might erode, so retailers may gradually
abandon the use of fi nal-digit nines.
9
However, after the Netherlands began requir-
ing retailers to round prices to the nearest zero or fi ve cents, retailers continued the
practice of ending most prices with nine [Tiplady, 2004].
144
EASTERN ECONOMIC JOURNAL
IMPACT ON SEIGNIORAGE
Lombra’s fi nal point is that eliminating the penny will cost the U.S. Mint sub-
stantial seigniorage each year. The Mint reported seigniorage from pennies of $24.6
million for fi scal 2001 but since the publication of Lombra’s article these fi gures have
fallen substantially – to $8.8 million in 2002, $1.0 million in 2003 and $4.6 million in
2004. However, it is not clear that these calculations are accurate because the Mint
allocates overhead costs across its coins and the portion allocated to pennies is a very
small fraction.
10
A more realistic share of overhead attributed to pennies would wipe
out seigniorage from them completely.
11
Recent events have substantially undermined the argument that the penny
should be maintained because of the profi t it generates for the Mint. By any account-
ing the Mint will show negative seigniorage from pennies for 2005 and 2006 – and it
is likely to do so well into the future. The price of zinc, which comprises 97.5 percent
of a penny, averaged about 35 cents per pound in 2002 and much of 2003, rose to 45
cents at the end of 2003, passed 60 cents in early 2005, then exploded to $1.50 per
pound in mid 2006 and $2.00 near the end of the year [Kitco Base Metals, 2006]. In
mid-November 2006 the value of the metal in a penny equaled 1.1 cents. Analysts at
the International Monetary Fund blame the wider surge in metals prices on strong
international demand, especially from China. Although, they foresee a fall in these
prices as the supply curve becomes more responsive, positive seigniorage on pennies
will only return if the cost of the metal content falls back to at least the level of early
2003, but based on estimates for more widely used metals, like copper and aluminum,
this seems unlikely to occur by the end of the decade [IMF, 2006].
In addition, the Mint’s seigniorage calculation omits part of the cost of producing
pennies. It counts only the cost of distributing pennies from the Mint to the Federal
Reserve. The Federal Reserve’s cost of distributing pennies across the nation is likely to
be several-fold higher than the Mint’s distribution cost ($2 million), but it isn’t counted
against the penny in the standard calculation of the government’s profi t from pennies.
Chande and Fisher [2003] estimate the entire distribution costs of Canadian pennies
to be several times higher than the revenue from selling pennies, calculating that in
2001 the cost of manufacturing and distributing the one-cent coin was 3.95 cents per
coin. These fi gures are likely to be fairly similar for the United States.
Finally, seigniorage (or losses) from the penny shouldn’t be viewed in isolation.
Eliminating the penny could be associated with the replacement of the dollar bill with
a dollar coin. If the Mint were to forego its small (or negative) seigniorage from the
penny, it could probably earn substantially more seigniorage from the dollar coin.
Overall annual budgetary savings – including production and processing costs, sei-
gniorage revenue, start-up, and advertising cost – of switching from a paper to metal
dollar have been estimated at more than $500 million [GAO, 2000].
CONCLUSIONS
Recently, N. Greg Mankiw [2006] listed eliminating the penny as one of seven New
Year’s resolutions that politicians should make: “This year I will vote to eliminate the
penny. The purpose of the monetary system is to facilitate exchange, but I have to
145
TIME TO ELIMINATE THE PENNY
acknowledge that the penny no longer serves that purpose. When people start leaving
a monetary unit at the cash register for the next customer, the unit is too small to be
useful. I know that some people will be upset when their favorite aphorisms become
anachronistic, but a nickel saved is also a nickel earned.”
Days later Mark W. Weller [2006], Executive Director of Americans for Common
Cents, rebutted Mankiw citing the contention that consumers would be hit with a
“rounding tax” totaling hundreds of millions of dollars annually and surveys that 70
percent of Americans want Congress to keep the penny in circulation, and that 80
percent believe merchants will use rounding as an excuse to hike prices. The primary
source backing the contentions of this group (and others) that removing pennies will
harm consumers is Lombra’s paper [Americans for Common Cents, 2002].
My fi ndings, however, suggest that rounding of current prices would not gener-
ate a “rounding tax.” Would merchants respond to the penny’s elimination by hiking
prices? Looking at current prices cannot answer this question conclusively, but strongly
suggests that merchants will not. Mankiw’s contention that the penny has become
useless is an important one. Ironically, since the half-penny was eliminated in 1857,
consumer prices have risen more than twenty-fold [Williamson, 2005], so that the
half-penny of 1857 is the equivalent of today’s dime, having a value of 11.3 cents in
2006 dollars. In 1857 the average nominal wage for common labor was $1.01 per day
[Margo, 2000, Table 3A.5]. Since the workday averaged about eleven hours, hourly
pay was about 9.2 cents. Thus, the half-penny equaled about 1/18
th
of the hourly pay.
The equivalent fraction of today’s federal minimum hourly wage is 23.6 cents. By this
metric, the half-penny was worth almost the same as a quarter in today’s economy. It
is paradoxical that few complained about the elimination of the half-penny in 1857,
yet there is so much concern about the loss of today’s penny – whose relative value is
only 1/10
th
to 1/20
th
as much.
My conclusion is that the U.S. Mint should stop producing pennies. The U.S. gov-
ernment is losing money on penny production and in light of my fi ndings, it appears
that very little would be lost from eliminating the penny from the U.S. economy and
more could be gained. However, individuals have the ability to use whatever they
wish as money. If they wanted to continue using pennies, they could do so using the
accumulated stock of pennies after the U.S. Mint discontinues minting the penny.
NOTES
I thank Charlie Sweigart, Ron Padget, and Susan Flynn for helping me obtain the data used in this
analysis.
1. In addition some states exempt food from sales tax but allow localities to levy taxes on food. Adding
these states to those completely exempting food from taxes boosts the percentage of the population
living in states without a food sales tax to as much as 79.2 percent.
2. The transactions took place between July 17 and July 23, 2005. I omitted a small percentage of the
data that were obviously anomalous – those for which pre-tax or post-tax amounts were less than zero
and for which the tax rate was negative or higher than plausible.
3. Notice that 54.1 percent of the transactions involved change being returned. This may imply that
Lombra’s lower bound estimate (50 percent) of the percent of transactions involving cash is closer to
the mark. Looking more broadly, beyond just convenience stores, a survey for the American Bankers
Association (Dove Consulting, 2005) estimates that cash is used in 33 percent of in-store purchases.
146
EASTERN ECONOMIC JOURNAL
4. Change is sometimes returned from debit/credit transactions, but the amounts almost universally are
whole dollars, so if some of these exist in the data set, they won’t affect the rounding amounts.
5. One reason that customers may gain slightly from rounding in this data set is that they occasionally
inadvertently miss their intended mark when pumping gas – for example, aiming to pump $10.00 but
inadvertently pumping $10.01 or $10.02. If the penny were eliminated, consumers would have more
of an incentive to do this, but gasoline retailers would certainly respond with strategies that make it
more diffi cult to accomplish – such as slightly boosting the price of gasoline.
6. In addition, almost none of the federal government’s payments are made using cash, so it won’t be
affected by the “rounding tax.”
7. Ironically, gasoline sellers do have many sales that end in exact dollar amounts – because customers
so often pump an exact dollar amount of gas – but the opportunity for employee theft due to this is
not considered to be a signifi cant problem.
8. Many Internet sites report a study by the National Association of Convenience Stores and the
Walgreen’s drug store chain, which found that handling pennies adds 2 to 2.5 seconds to each cash
transaction. I have been unable to locate this study, however.
9. Interestingly, the nation’s largest retailer, Wal-Mart, has already moved away from this strategy.
10. For example, in fi scal 2004 the Mint’s total “selling, general and administrative” costs equaled $88.9
million. 64.8 percent of this was allocated to quarters, 28.8 percent to dimes, 0.9 percent to nickels,
and only 1.1 percent to pennies, despite the fact that pennies represented 22.3 percent of cost of goods
sold and 3.2 times as many pennies were produced as quarters.
11. Likewise, the Mint appears to have recently redefi ned its cost breakdown to make the penny look more
profi table than it is. In fi scal 2001, when the penny was more profi table, the Mint counted “general
and administrative” costs for pennies as $3.7 million. In fi scal 2004, it counted “selling, general and
administrative” costs for pennies as only $1.0 million – as the quantity of pennies sold fell by 44 percent,
its overhead costs were calculated to have fallen by 73 percent. In fact, overhead costs assigned to
pennies fell by more than 50 percent from 2003 to 2004, although revenue from pennies increased.
REFERENCES
Americans for Common Cents. Penny Research. http://www.pennies.org/research.html, accessed
February 23, 2006.
Bureau of Labor Statistics. National Compensation Survey: Occupational Wages in the United States,
July 2004. Washington, DC, 2005.
Chande, D. and Fisher, T. Have a Penny? Need a Penny? Eliminating the One-Cent Coin from Circula-
tion. Canadian Public Policy, 2003, 511-17.
Dove Consulting. Consumer Payment Preference: Understanding Choice. www.aciworldwide.com/pdfs/
consumer_payment_preferences_trend.pdf, 2005.
Federation of Tax Administrators. State Sales Tax Rates. http://www.taxadmin.org/fta/rate/sales.html,
accessed February 23, 2006.
General Accounting Offi ce. Financial Impact of Issuing the New $1 Coin. Washington, DC, 2000.
International Monetary Fund. The Boom in Nonfuel Commodity Prices: Can It Last? World Economic
Outlook, September 2006, Chapter 5.
Kitco Base Metals. http://www.kitcometals.com/charts/zinc_historical.html, accessed November 15, 2006.
Lombra, R. Eliminating the Penny from the U.S. Coinage System: An Economic Analysis. Eastern Eco-
nomic Journal, 2001, 433-42.
Lotz, S. and Rocheteau, G. The Fate of One-Dollar Coins in the U.S. Federal Reserve Bank of Cleveland,
October 15, 2004.
Mankiw, N. G. Repeat after Me. Wall Street Journal, January 3, 2006.
Margo, R. Wages and Labor Markets in the United States, 1820-1860. Chicago: University of Chicago
Press, 2000.
Tiplady, R. Small Change, Big Annoyance in Europe. Business Week Online. September 23, 2004.
United States Mint. Annual Report. Washington, DC, 2002 and 2004, http://www.usmint.gov/about_the_
mint/index.cfm?action=annual_report
Weller M. A Penny for Your Economy. Wall Street Journal, January 11, 2006, A15.
Williamson, S. What Is the Relative Value? Economic History Services, December 14 2005, http://eh.net/
hmit/compare/, accessed February 23, 2006.

Discussion

Tap the paper where you want to comment
More detailed data about the number of transactions in a typical basket and the total transaction sum (i.e. if you buy three 99 cents items, you would round down the total), and a more detailed analysis of the impact of sales tax on the basket sum (if rounding happens after items are taxed, totals should be more random and uniformly distributed between 0s, 1s, 2s, 3s, 4s...etc) are needed in the original papers' analysis.