The free-play tax deduction debate: How academic research can help

UNLV Gaming Research & Review Journal t Volume 21 Issue 1

The Free-Play Tax Deduction

Debate: How Academic Research

Can Help

Anthony F. Lucas

Katherine Spilde

Introduction

Free-play has become ubiquitous in casinos, with operators claiming its purpose

ranges from extending play time (Gruetze, 2012), to lowering the house advantage on slot

machines (Burns, 2010), to competing for market share (Murphy, 2016), to increasing

customer loyalty (Armon, 2015), to growing net revenues (Armon, 2015; Belko, 2016;

Burns, 2010; Gruetze, 2012), and more. As gaming proliferates across the globe,

many operators have increased their reliance on free-play oers, to combat increased

competition. For example, in one U.S. jurisdiction, it is not uncommon for an operator to

redeem in excess of $50 million dollars of free-play per year (Belko, 2016). For others,

free-play redemptions comprise 20% or more of a casino’s total gaming revenues (Barker,

2015; Murphy, 2016).

This increased reliance on free-play has caused several U.S. jurisdictions

to rethink the tax treatment of these incentives (Belko, 2016, Armon, 2015). Other

tax authorities are interpreting existing and arguably vague regulations, with great

consequence to gaming operators (Brunt, 2016). With many governments suering

revenue shortfalls, the free-play tax credit will likely remain a contentious issue.

In short, operators want to deduct the full face value of all free-play oers

redeemed, but taxing authorities are questioning the assumptions that underlie the

rationale for this tax treatment. The positions adopted by both sides are anchored in large

part by claims, assumptions and questions related to the ecacy of free-play oers.

In spite of this, we were unable to locate any references by either side to the academic

research that has addressed this very issue, i.e., the ecacy of free-play oers. The

primary aim of this work is to introduce the results of these research papers into the

free-play tax treatment debate. Such results oer an alternative, objective and empirical

means of evaluating key points that are central to both arguments. Going forward, the

methodologies and results from these studies oer a means to spark new and meaningful

discussions, reshape the free-play debate, and inuence the formation of future tax

policies.

This paper begins by describing the mechanics, evolution, critical issues and

assumptions associated with free-play oers. Next, we address the basic structure of

gross gaming revenue taxes, followed by a review of existing free-play tax policies, and

the arguments for the dierent tax treatments associated with these oers. Finally, we

demonstrate the extent to which each argument is supported by the empirical academic

research, leading to recommendations for mutually benecial policy revision.

Anthony F. Lucas

Professor

William F. Harrah College

of Hotel Administration

University of Nevada,

Las Vegas

[email protected]

(702) 300-6064

Katherine Spilde

Associate Professor

Chair, Sycuan Institute on

Tribal Gaming

L. Robert Payne School of

Hospitality &

Tourism Management

San Diego State University

[email protected]

(760) 533-9387

UNLV Gaming Research & Review Journal t Volume 21 Issue 1

Free-play Oers

There are two types of free-play oers: discretionary (DFP) and earned (EFP).

Both forms are essentially electronic currency which must be wagered before gamblers

can claim any remaining/surviving credits from the face value of the original award

(Fine, 2009; Lucas & Spilde, 2017). For example, a gambler redeeming an oer with a

face value of $20 must place at least $20 dollars of wagers before she would be permitted

to cash-out any surviving free-play credits from the original $20 award. That is, after

placing $20 in wagers, she may have a credit balance ranging from zero to a potentially

great sum. The upper limit of the surviving credit balance is dened in part by the top

award of the game on which the oer is redeemed.

Background

In the mid-1990s, Las Vegas casinos catering to the local market initiated large-

scale cash mail programs designed to create additional visits and increase spend-per-trip

(Lucas & Kilby, 2008). These oers were a precursor to the modern free-play oers.

Following the cash mail campaigns, many slot clubs oered members a popular benet

known as same-day cash back. This was a more immediate version of the same benet.

Both cash mail and same-day cash back were essentially partial refunds of the casino’s

theoretical win earned from each player. The former required a direct mail piece, whereas

the latter utilized onsite redemption technology.

One common problem with both of these early rebate incentives was known as

the walkout phenomenon. Lucas & Kilby (2008) described an observation study in which

45% of patrons redeeming cash mail coupons walked out of the subject casino without

gambling. The mere fact that players could leave without gambling bothered many

operators, which led to the development of technology that required the recipients of

these play incentives to wager them at least once before cashing-out any surviving credits

(Fine, 2009; Gruetze, 2012).

The disappointing results produced by early time series regression models

designed to measure the impact of cash mail programs were likely explained away by the

walkout phenomenon (Lucas & Kilby, 2008). Although the new free-play technology was

touted as a cure for the walkout problem (Fine, 2009), questions remained regarding the

original purposes of these incentives. Specically, do they generate increased spend-per-

trip, and if so, is it enough to cover the oer costs? Also, do they produce incremental

gaming trips? In spite of the broad industry support for free-play, some gaming insiders

question the unconditional ecacy of these oers (Fine, 2009; Gruetze, 2012; Lucas &

Kilby, 2008; Murphy, 2016; Sortel, 2010).

Chief among these concerns is play time. For example, if operators are too

generous with free-play awards and players have limits on the time allotted for gambling,

it can take too long to lose the free-play credits. The issue here is that players do not

access much of their own bankroll if any at all, because it takes too long to lose the

free-play awards (Fine, 2009; Gruetze, 2012; Murphy, 2016; Sortel, 2010). What good

is an extra trip, if it is only to redeem credits awarded by the casino? Additionally, how

is spend-per-trip increased by wagers stemming primarily or solely from casino-issued

credits? These are fair questions, given the severity of the recent uptick in free-play issues

(Belko, 2016; Kollars & Otte, 2015; Murphy, 2016).

Recent campaigns have grown to unexpected levels (Belko, 2016; Kollars &

Otte; 2015), creating an unanticipated yet pressing taxation issue. Further, some gaming

regulations and revenue sharing agreements do not specically, clearly or adequately

address the tax treatment for free-play (Burns, 2010; Brunt, 2015), as it was not a critical

issue at the time they were created.

UNLV Gaming Research & Review Journal t Volume 21 Issue 1

Free-Play Tax Deduction

Gross Gaming Revenue Taxes

There is no single denition for how gaming taxes are computed. There are

dierences across jurisdictions and computational variations/options within jurisdictions

(Lucas & Kilby, 2012). For example, in Nevada, operators are permitted to pay monthly

gaming taxes on either a cash basis or an accrual basis (Lucas & Kilby, 2012). These are

essentially top-line or revenue taxes, similar to the basic notion of a sales tax. They are

most often referred to as gross gaming revenue (GGR) taxes, or revenue sharing payments

within the tribal gaming context. In general, casinos pay a GGR tax on the dierence

between (1) the dollar value of wagers won by the casino; and (2) the dollar-value of the

payouts on wagers lost by the casino. Alternatively stated, the start position for GGR

liability in most jurisdictions is the dierence between the dollar-amount lost by losing

players and the dollar-amount won by winning players. From this basic start position,

items such as free-play redemptions are addressed. More specically, in Nevada, for

example, NRS 463.0161 provides the following base calculation for what is known as

GGR:

“Gross revenue” means the total of all:

(a) Cash received as winnings;

(b) Cash received in payment for credit extended by a licensee to a patron for

purposes of gaming; and

not party to a wager, less the total of all cash paid out as losses to patrons, those

amounts paid to fund periodic payments and any other items made deductible

as losses by NRS 463.3715. For the purposes of this section, cash or the value

of noncash prizes awarded to patrons in a contest or tournament are not losses,

except that losses in a contest or tournament conducted in conjunction with an

inter-casino linked system may be deducted to the extent of the compensation

received for the right to participate in that contest or tournament.

In Nevada, gross gaming revenue does not include “Any portion of the face

value of any chip, token or other representative of value won by a licensee from a patron

for which the licensee can demonstrate that it or its aliate has not received cash;”

or “Cash provided by the licensee to a patron and subsequently won by the licensee,

for which the licensee can demonstrate that it or its aliate has not been reimbursed.”

(NRS.0161, Part 2, Subparts c & g). This language removes the face value of redeemed

free-play oers from the monthly gross revenue calculation. Although beyond the scope

of this paper, NRS.0161, Part 2 also excludes other items from the calculation of gross

revenue.

Current Free-play Tax Policies

Policies and regulations related to the deductibility of free-play oers within the

gaming tax and revenue sharing calculations are not consistent across jurisdictions. Some

taxing authorities allow operators to reduce their taxable/sharable win by the face value

of the free-play oer, while others limit or cap such deductions by a percentage of win or

a at dollar amount (Armon, 2015; Spectrum, 2014). Some do not allow any deductions

for free-play (American Gaming Association, 2015), and others are considering new

limitations (Belko, 2016; Armon, 2015). See Table 1 for an abridged schedule of free-play

deductibility by U.S. jurisdiction.

UNLV Gaming Research & Review Journal t Volume 21 Issue 1

Table 1

Sample of Free-play Tax Policies Among U.S. Jurisdictions: Commercial Casinos

Jurisdiction Free-play Policy*

Delaware Permitted to issue 20% of prior year’s net terminal income on a

tax free basis.

Florida Value of free-play not treated as net win for slot machines.

Indiana Value of free-play over $5 million at a single property is treated

as revenue.

Louisiana Free-play taxed as normal revenue.

Maine Cash prizes, winnings or credits from promotional credits are

considered gross slot machine income.

Maryland 20% of total VLT income from previous scal year may be used

as tax free promotional play.

Massachusetts Value of free-play not included in gross revenue calculations.

Michigan Value of free-play treated as revenue.

Mississippi Free-play only taxed if the promotional credits have a cash

value.

Missouri Value of free-play treated as revenue.

Nevada

Value of free-play not treated as revenue.

New Jersey First $90 million of free-play taxed as gross revenue.

New Mexico Value of the free-play treated as revenue (within racinos).

New York Maximum of 15% of a facility’s net win is excluded from

revenue calculations.

Ohio Free-play not treated as revenue.

Pennsylvania Value of free-play not treated as revenue.

Rhode Island 10% of the previous year’s Net Terminal Income plus $750,000

may be used as tax free promotional play.

West Virginia Permitted tax free issues of between 2% and 3% of net revenue.

Note. Adapted from the American Gaming Association (2015)

* In some cases, free-play tax policies for tribal casinos vary from those applied to commercial casinos

within the same jurisdiction.

UNLV Gaming Research & Review Journal t Volume 21 Issue 1

Table 1 demonstrates stark dierences in free-play tax polices even among jurisdictions

that share a geo-political boundary. Such dierences suggest (1) considerable uncertainty

and dierence of opinion regarding the incremental impact of free-play oers; and (2) a

need for clarication of these eects. The next two sections of this review describe the

rationale behind these considerably dierent policy positions.

Deductibility Arguments

Casino operators are the primary proponents of this argument, which rests on (1)

the unfairness of disallowing free-play deductions; and (2) that free-play expands the tax

base by generating incremental wagering activity (Belko, 2016; Burns, 2010; Murphy,

2016). The second of these two claims can be further unpacked in two parts. First, the

tax base is expanded by increased spend-per-trip (Murphy, 2016). Second, the tax base is

increased by creating incremental visits (Spectrum, 2014).

The phantom revenue concept underlies the argument for the unfairness of

disallowing free-play deductions (Burns, 2010; Cantrell, n.d.). For example, let’s assume

a player (1) wagers $20 in free-play in a slot machine; (2) eventually loses all $20 to the

game; and (3) the game receives no other wagers on that gaming day. Regardless of the

precise dollar amounts, the coin-in for this game would be $20 greater than the coin-out,

resulting in a casino “win” of $20. The operators would claim that this $20 is phantom

revenue. Specically, if they were to include the face value of the free-play coupon in

coin-in, they would be paying taxes on $20 of win that they never collected.

Expanding on the previous example, let’s assume that $500 of wagers (i.e.,

coin-in) were generated in the course of losing the $20 of free-play. This means that all

forms of payouts on this game would equal $480, creating a casino win of $20 (i.e., $500

- $480). Alternatively, if operators were permitted to reduce the coin-in by the dollar

amount of its face value (i.e., $20), then the tax liability would reect the operator’s

true gain. For example, $500 - $20 - $480 = $0, which reects the true change in the

operator’s cash position. All of this is based on a popular assumption among operators,

holding that free-play redeemers wager their free-play credits until their award balance is

zero (Fine, 2009).

Others have argued “unfairness” by way of examples from other industries,

citing the absence of coupons and loyalty club benets from the sales tax base in grocery

stores (Armon, 2015; Burns, 2010). Additional examples include nontaxable benets

from hotel, airline, and rental car loyalty clubs, such as earned room nights, tickets, and

upgrades/rentals. In all of these cases, the face value of the redeemed benets is not

included in the calculation of taxable sales. Deductibility proponents argue that based on

precedent established in other industries, free-play should be aorded the same treatment.

Next, we will describe the rationale for expanding the tax base. From Rüdisser,

Flepp and Franck (2015), European casino operators described the rationale for awarding

free-play in terms of what Thaler and Johnson (1990) dubbed the house money eect.

In general, Thaler and Johnson found that investors who experienced windfall gains

were less risk averse in subsequent transactions, vis-à-vis those who did not experience

windfall gains. Similarly, the casino operators held that staking gamblers with free-play

would lead to riskier wagering behavior, which would then carry over into subsequent

betting activity, once the free-play credits were exhausted. Ultimately, it is assumed that

the casino’s win will be increased by the free-play redemption process. In this context,

the operators are assuming that free-play redemption will increase the spend-per-trip

by way of wagers placed with the player’s own bankroll (Burns, 2010; Gruetze, 2012;

Sortel, 2010).

Free-Play Tax Deduction

UNLV Gaming Research & Review Journal t Volume 21 Issue 1

Another argument for expanding the tax base via free-play is that the oers

generate incremental gaming trips (Spectrum, 2014). Of course, more trips equal more

win, assuming the win per trip remains at a level sucient to cover the free-play oer

costs. The idea here is that the free-play oers create loyal customers who then visit the

casino more often.

Finally, there are concerns that policy restrictions related to the deductibility of

free-play oers could diminish the ability of operators to compete across jurisdictional

boundaries (Belko, 2016). For example, if Pennsylvania were to enact regulations that

discouraged free-play oers, then operators in that state would suer a competitive

disadvantage against casinos in neighboring states with presumably less restrictive

free-play regulations. This argument appeals to the common concern for citizens in the

home state spending money in a neighboring state (Armon, 2015; Kollars & Otte, 2015;

Murphy, 2016). The political and economic eects of such behavior are unappealing to

most state governments.

No-Deductibility Arguments

The phantom revenue argument is countered by challenging the critical

assumption that the player continues to place wagers until all free-play credits are lost.

Alternatively, the players could elect to pocket the credits that survive the compulsory/

minimum wagering requirement (Fine, 2009), becoming a costly free rider (Cigliano,

2000). For example, let’s assume a player redeemed $20 in free-play credits on a slot

machine with a 10% casino advantage. On average, the player would have $18 in credits

after satisfying the one-time compulsory play requirement (i.e., $20 – ($20 x 10%) =

$18). At this point, the player could cash-out the $18 and leave the casino.

If this occurs, and the operator is permitted to reduce the coin-in by the face

value of the free-play award, the taxable win would be under-reported. For example,

$20 coin-in - $20 free-play - $18 payouts = $18 loss. In this case, the wagering activity

associated with free-play redemption would produce a loss of $18 for gaming tax

purposes. Taxing authorities argue that such a loss is created by operators reclassifying

promotional expenses as contra-revenue items. Further, legislators have noted that the

taxpayers should not have to foot the bill for casino promotions (Armon, 2015).

This argument has been expanded to address the nature and origin of free-

play credits. Specically, these credits/awards do not meet the denition of a prize (i.e.,

payout) that is directly related to a wager. They are referred to by many in the industry

as a reinvestment expense (Barker, 2015; Fine, 2009). For example, slot players are not

winning discretionary free-play awards as they would a programmed payout resulting

from a winning combination of symbols. Free-play rewards are a conscious post hoc

investment in players (i.e., a marketing expense), as opposed to a payout associated with

a wager. Because of this dierence, they should not be deducted from overall coin-in, for

the purposes of computing taxable gaming win.

Slot accounting systems cannot dierentiate between payouts resulting from

free-play wagers and cash wagers (Lucas & Spilde, 2017; Robison, 2014). These systems

were simply not designed to do this, and the appropriate technological adjustments/

accommodations do not appear likely, for reasons beyond the scope of this paper. Because

of this technological limitation, all payouts are deducted in the gaming tax formula,

regardless of the source of the payout (i.e., a free-play or a cash wager). Some gaming

regulators contend that if the payouts associated with free-play wagers are to be included,

then it is only fair that the wagers that produced these payouts are also included (Brunt,

2015). Failure to do so would result in articially low gaming win. As regulators watch

free-play redemptions soar in jurisdictions with operator-friendly policies, limiting the

deductibility of these play incentives appears increasingly lucrative to state governments

(Armon, 2015; Belko, 2016).

UNLV Gaming Research & Review Journal t Volume 21 Issue 1

Next, we examine the other side of the deductibility argument related to the

omission of promotional sales from the sales tax calculation. This argument is countered

by highlighting a critical dierence in the free-play transactions. Specically, free-play

redemptions can result in decreased gaming win (i.e., taxable revenue), whereas coupon

or promotional awards in the grocery, hotel, airline and rental car industries cannot

decrease revenue that has already been collected. That is, these transactions cannot lower

the existing revenue level. After all, regulators have a responsibility to protect the tax

base, and all free-play payouts are funded from that shared till.

In both Macao and New Jersey, when casino operators issue credit to players

and those players lose any dollar amount of the advanced funds, a gaming tax liability

is created, regardless of whether this loss is collected by the casino operator. Even

when no money is collected, the gaming tax is still owed. Such policies were designed

to encourage operators to be careful, responsible and diligent when issuing credit. And

taxpayers do not bear any part of the burden of unsuccessful collection eorts.

While credit-collection and free-play tax policies are not precisely equivalent

issues, some have made the general argument that policy encourages practice (Belko,

2016). If restrictive credit policies prevent capricious credit granting practices, then

restrictive free-play policies may prevent excessive free-play issues. Without robust

empirical evidence that free-play is expanding the gaming tax base, should taxing

authorities be expected to take operators at their word? After all, several academic studies

have described failed casino marketing tactics associated with similar claims, most of

which were aorded and/or encouraged by favorable gaming tax treatment (Lucas, Kilby

& Santos, 2002; Lucas, 2004; Lucas & Bowen, 2002). The next section examines the

academic literature to clarify what is known about the relationship between free-play

redemption and incremental gaming activity.

Academic Research Results

There are only a few studies that have examined the impact of free-play oers

on rated, slot wagering behavior. Further, none have attempted to measure the creation

of additional trips spurred by free-play oers. It is important to remember that measuring

the impact of free-play campaigns is dicult, as many concomitant sources of inuence

aect casino business volumes. The challenge is to estimate the unique eect of the

free-play oers within this multidimensional space. Such estimates are critical, as they

represent the foundation of the deductibility arguments. The following paragraphs

describe the extant academic studies in chronological order.

Lucas, Dunn and Singh (2005) examined data gathered from 2000 through

2003 from an integrated resort on the Las Vegas Strip. This time period featured both the

opening of the casino and the subsequent implementation of free-play oers, providing

trip-level data before and after the free-play campaign. The design featured two groups:

(1) slot players who made gaming trips without free-play incentives; and (2) the same

players who made trips with $50 free-play incentives. The same design was repeated for

players who received $100 free-play awards. Their study sought to measure the impact of

the free-play oers by measuring the change in rated, trip-level, wagering volume across

the two oer conditions (i.e., no-oer trips vs. trips with the free-play oer). Rated coin-

in served as the criterion variable, as players activated free-play awards via their loyalty

club cards.

Multiple regression analysis was employed to estimate the impact produced

by the binary free-play variables. The eects of these variables were examined within

a model that included several covariates identied as alternative sources of inuence

on trip-level wagering volume. In the $50 data set, the free-play variable produced a

signicant and negative eect (B = -$355.98, p < 0.05). This result indicated that the

Free-Play Tax Deduction

UNLV Gaming Research & Review Journal t Volume 21 Issue 1

free-play trips were associated with a $355.98 decline in wagering volume from the level

of the no-free-play trips. In the $100 data set, the free-play variable also generated a

negative model eect, but it was not statistically signicant (B = -236.78, p > 0.10).

The authors noted that the results were consistent with the idea of bankroll

cannibalization. This observation was based on the nding that the free-play trips

produced less wagering volume than the no-free-play trips. Given that the same players

produced these trips, it appeared that they were possibly substituting the free-play awards

for a portion of their own bankrolls. These results are consistent with the previously

mentioned fears that players limit their gambling time. Specically, the time required to

lose all of the free-play credits cuts into the time the casino has to win a player’s own

bankroll.

Suh (2012) analyzed data gathered in 2007 from a major Las Vegas Strip

hotel-casino resort. Her work consisted of a eld experiment in which slot players from

a common tier of the casino’s data base were divided into two groups. Under normal

circumstances, all of the subjects would have received a $50 free-play award; however,

for the experiment, half of the participants received the normal $50 award and half

received a $100 award. The objective was to determine whether the increased free-play

award would lead to a greater spend-per-trip (i.e., rated coin-in per trip).

It was not surprising that the redemption rate for the $100 free-play awards was

much greater. In fact, it was 45% greater than the redemption rate for the $50 awards.

This led to a nal sample comprised of 155 subjects from the $50 group and 225 subjects

from the $100 group.

Multiple regression analysis was used to test a theoretical model similar to

that advanced in Lucas et al. (2005). The results associated with the categorical free-

play variable indicated that the increase in the free-play award value from $50 to $100

failed to signicantly increase the rated, trip level, wagering volume of the subjects

(B = $617.25, p > 0.05). Like Lucas et al. (2005), Suh (2012) discussed the possibility

of bankroll cannibalization as it related to her result. She also echoed Lucas et al.

(2005) with respect to potential entitlement eects stemming from protracted free-play

campaigns. That is, players had grown accustomed to receiving free-play oers on a

regular basis and began to feel entitled to these oers, with no obligation to behave any

dierently.

Suh, Dang and Alhaery (2014) analyzed data gathered in 2006 from a U.S.

riverboat casino. They used time series regression analysis to test a theoretical model

designed to explain the variation in daily rated coin-in. One of the predictor variables

represented the daily dollar value of all redeemed free-play oers. The clientele of

the casino was ideal for the analysis of free-play ecacy, as it was chiey comprised

of frequent visitors. With numerous opportunities for redemption, free-play oers are

thought to be particularly attractive to frequent visitors (Lucas & Kilby, 2008).

The framework of the theoretical model employed by Suh et al. (2014) was well

established; however, estimating the eects of free-play was not the primary purpose

of their study. They simultaneously examined the eects of several forms of casino

promotions. The covariates of their model included the following variables: Days of the

week, holidays, special event days, and several other casino promotion variables. The

model explained 94% of the variation in daily, rated coin-in (F = 220.20, df = 17 & 223,

p < 0.0005), but the variance ination factors ranged from 1.0 to 5.9. The authors were

careful to note the cautionary level of multicollinearity, especially when interpreting the

eects of individual predictor variables.

UNLV Gaming Research & Review Journal t Volume 21 Issue 1

The free-play variable did produce a statistically signicant eect (B = $10.90, p

< 0.0005). For every one dollar of free-play redeemed, rated coin-in could be expected to

increase by $10.90, ceteris paribus. But coin-in is not win. To compute the expected win

(or revenue), coin-in must be multiplied by the casino’s average advantage on all wagers.

Although not provided by the authors, a reasonable estimate of this advantage would be

7.5%. In this case, for every one dollar of free-play redeemed, the casino could expect to

win $0.82 (i.e., $10.90 x 7.5%). Under these assumed conditions, the results suggested

that the free-play campaign did not expand the tax base.

The value of the free-play coecient increased dramatically when the same

model featured total coin-in as the criterion variable. A second riverboat model also

produced a positive and signicant free-play coecient of even greater magnitude, with

total coin-in serving as the criterion variable. However, total coin-in included wagers

placed by patrons other than the slot club members (i.e., it included unrated play). Slot

club cards must be inserted to redeem free-play rewards and future free-play awards

are earned by allowing the casino to track a patron’s play. For these reasons, rated coin-

in is the preferred dependent variable when estimating the ecacy of free-play oers.

Again, Suh et al. (2014) set out to examine the eects of many dierent forms of casino

promotions, so total coin-in was an appropriate criterion variable for their models. Their

primary concern was not limited to estimating the eects of free-play.

Rüdisser et al. (2015) analyzed data from a 2015 eld experiment conducted

in a Swiss casino. Their experiment was designed to measure the gambling behavior

associated with free-play awards, with specic attention to the level of risk aversion

demonstrated by redeemers. Unlike free-play distributed within an ongoing campaign,

participants arrived at the casino unaware of their chance to win free-play awards. Upon

arrival, patrons were randomly selected to spin a prize wheel with random outcomes. The

result of the spin determined the value of the free-play awards, which included zero.

Based on the outcomes from the prize wheel, a control group was established

with patrons who received no free-play oer, while the treatment group was populated

by those who won free-play awards. These treatment group awards ranged from CHF 5

to CHF 50 in value, as all monetary data were expressed in terms of Swiss Francs (i.e.,

CHF). At the time of their study, Swiss Francs were very nearly equivalent to U.S. dollars.

Although the authors made several comparisons of outcomes produced by the

control and treatment group members, the general conclusion was that those who received

free-play awards recorded smaller average losses than those who did not. The casino’s

average win-per-visit from gamblers in the treatment group was CHF 153, while the same

for the control group was CHF 263. Similarly, the casino’s average theoretical win was

CHF 105 for members of the treatment group and CHF 207 for members of the control

group. All outcomes were measured by way of tracking cards, i.e., rated play.

The authors concluded that casino operators should not issue free-play oers,

as they were both costly in terms of face value and they were associated with smaller

average losses per gambler. The authors also noted that the ndings supported a

phenomenon known as loss aversion, from behavioral economics (Tversky & Kahneman,

1991, 1992). In general, loss aversion contends that people suer losses to a greater extent

than they enjoy gains of the same magnitude. In Rüdisser et al (2015), the free-play award

served as an endowment, or basis for loss. Once endowed, these patrons demonstrated

greater risk aversion than the control group in terms of wagering behavior. Further

support for this theory was provided by the dierence between the average bet for the

treatment group (CHF 7.71) and the control group (CHF 18.57). The authors noted that

the lower average bet for the treatment group was evidence of risk/loss aversion at work.

Free-Play Tax Deduction

UNLV Gaming Research & Review Journal t Volume 21 Issue 1

As previously mentioned, the casino marketers were assuming that the free-play

oers would illicit the house money eect, as described in Thaler and Johnson (1990).

To the contrary, the results failed to support this critical assumption. Given the abundant

claims of tax base expansion, this is a particularly noteworthy result.

Lucas and Spilde (2017) examined free-play redemptions and rated coin-

in data gathered from two tribal casinos referred to as Resort A and Resort B. Their

data sets were gathered in 2014 and were each comprised of 365 sequentially ordered

daily observations. The primary goal of their research was to estimate the impact of

discretionary free-play redemptions on rated coin-in levels. This question is closely

related to the ability of free-play to expand the tax base.

The two resorts diered importantly in terms of their scale of operation and

free-play campaigns. Resort A’s annual rated slot win was estimated at $133 million,

while the same for Resort B was $14 million. Resort A redeemed $16 million in

discretionary free-play oers over the course of the 365-day sample period, while Resort

B redeemed only $1.3 million in discretionary free-play over the same time period.

Their ndings were mixed in that Resort A’s results were not consistent with

the idea of expanding the tax base, while Resort B’s results supported the notion of

expanding the tax base via free-play. The time series models produced signicant

and positive free-play coecients for both resorts, but the criterion variables did not

represent win. For Resort A, a one dollar increase in the free-play variable produced an

$11.75 increase in rated coin-in (p < 0.0005), while a one dollar increase in Resort B’s

free-play variable generated a $24.27 increase in rated coin-in (p < 0.0005).

These coecients had to be converted into incremental win estimates to address

issues such as tax base expansion. This was achieved by multiplying the coecients by

each resort’s expected win percentage on slot machine wagers. Ultimately, Resort A was

redeeming one dollar of free-play in exchange for $0.88 of incremental win, suggesting

the creation of a negative cash ow. Resort B fared better, swapping a dollar of free-play

for $1.64 of incremental win. Even after the associated variable costs of issuing and

redeeming the oers, Resort B looked to reap considerable benets from its free-play

campaign, while simultaneously expanding the tax base. Unfortunately, Resort B is the

only property to post this kind of result. The bulk of the previously reviewed results

suggest the contrary, with respect to rated play (Lucas et al., 2005; Rüdisser et al., 2015;

Suh, 2012; Suh et al., 2014).

Related Casino Marketing Tactics

As casino gambling is comparably new to most jurisdictions outside of

Nevada, it is natural to look to Nevada for regulatory guidance. This approach can

be problematic, in spite of Nevada’s considerable regulatory experience and success.

Nevada’s regulatory and economic perspectives are somewhat dierent from those of

many other jurisdictions. Nevada sought to build an economy around the gambling

industry, as opposed to allowing limited exposure to it (Kilby, Fox & Lucas, 2004).

Many states seem to have legalized gambling to avoid raising taxes, stem short-term

budget shortfalls and prevent citizens from fueling the economy of neighboring states

(Barker, 2015; Belko, 2016; Murphy, 2016). This may be why Nevada has one of the

lowest gaming tax rates of any major jurisdiction (Pollack, 2010; American Gaming

Association, 2015). It does not seek to restrict the number of potential licensees and/or

limit the growth of the industry.

How does all of this relate to the issue at hand? Nevada’s low tax rate and

favorable tax treatment of marketing incentives may both stem from the strategy of

creating a growth-friendly regulatory environment. For example, unlike New Jersey and

Macao, Nevada operators do not pay taxes on uncollectible gaming debts (Inside Asian

UNLV Gaming Research & Review Journal t Volume 21 Issue 1

Gaming, 2011; NRS, 1997b; The Analyst, 2016; Thompson, 2015). Additionally, they

receive tax breaks on match-play incentives, free-play redemptions, promotional chips,

and discounts on loss (NRS, 1997b). For jurisdictions that do restrict the number of

licensees and seek to limit the growth of the gaming industry, this may not be the ideal

tax treatment for these incentives. For that matter, it may not be the ideal treatment for

Nevada. The following paragraphs make a case for questioning the ecacy of several

popular play incentives and gaming promotions.

Lucas (2004) estimated the impact of match-play redemptions on the cash

drop of blackjack table games in a Las Vegas Strip casino. The stated purpose of the

match-play campaign was to bring new players to the blackjack games. Alternatively

stated, the purpose was to grow blackjack revenues by way of a match-play incentive.

The results of his time series regression analysis indicated that a one dollar increase in

match-play redemptions resulted in an $8.99 decrease in blackjack cash drop. Similar to

Lucas et al. (2005) and Suh (2012), the author cited potential bankroll cannibalization

as a possible explanation for the unfavorable result. This result does not support the

arguments made for the preferential tax treatment of match-play coupons (Barker, 2015;

Burns, 2010).

Binkley (2001) and Lucas et al. (2002) chronicled the monumental damage to

overall table game protability caused by discounts on loss, also known as rebates on

loss. Additionally, Lucas & Bowen (2002) highlight the disappointing results associated

with the ubiquitous lottery promotions conducted by casino marketers. Specically, the

estimated incremental revenue gains fell well short of the incremental costs for three

large-scale lottery promotions. The stated goal of the lottery promotions was to drive

short-term protability in the casino via increased slot play.

Additionally, many casino marketers now oer at least a portion of the

guaranteed prizes in the form of free-play, as opposed to cash (Stations Casinos,

2017; Seminole Hard Rock, 2017). The guaranteed prize structure of a major lottery

promotion can reach levels of $500,000 or more, over longer events (e.g., 2 to 5 weeks),

and exceed $100,000 for a single-day event (Casino Arizona, 2017; SCA Promotions,

2017; Stations Casinos, 2017; Seminole Hard Rock, 2017).

Just because Nevada regulations allow for liberal tax treatment of gaming

incentives, or operators claim that promotions are creating incremental prots, does

not mean that either of these positions is valid. To the contrary, there is an abundance

of empirical evidence that suggests casino marketers are missing the mark with certain

play incentives and promotions. It is therefore reasonable for taxing authorities in

newer jurisdictions to question the tax treatment of incentives such as free-play. Maybe

policies and regulations should not encourage or subsidize practices that have been

found to decrease cash revenues and/or operating prots. To expand on Gu (2003),

when it comes to free-play, maybe operators are buying revenues at the expense of

prots.

Discussion

The results from the bulk of the reviewed academic research failed to support

the casino operators’ argument that free-play expands the gaming tax base (Lucas,

2004; Lucas et al., 2005; Rüdisser et al., 2015; Suh, 2012; Suh et al. 2014). Further,

several studies suggested that redeemers often cash-out at least some of the free-

play credits that survive the compulsory play requirement (Lucas & Spilde, 2017;

Rüdisser et al., 2015; Suh et al., 2014). These ndings are consistent with concerns

related to the time it takes for the casino to win excessive free-play awards (Fine,

2009; Gruetze, 2012; Murphy, 2016; Sortel, 2010), and the existence of costly free

riders among the redeemers (Cigliano, 2000). Several other studies support the idea

Free-Play Tax Deduction

UNLV Gaming Research & Review Journal t Volume 21 Issue 1

of bankroll cannibalization (Lucas, 2004; Lucas et al., 2005; Suh 2012). Finally, only

Lucas and Spilde (2017) oered partial support for Taylor and Johnson’s house money

eect, while the ndings from multiple studies at least partially supported Tversky

and Kahneman’s notion of loss aversion (Lucas et al., 2005; Lucas & Spilde, 2017;

Rüdisser et al., 2015). Given these results, it may be dicult for taxing authorities

to take operators and industry pundits at their word, regarding the net revenue

contributions of free-play. Of course, these alleged contributions are critical to the

argument for full face-value tax deductions. Overall, the results failed to support the

claim that the face value of free-play awards is fully recovered by the casino. By

extension, it would be dicult to argue that the results provided compelling evidence

of tax base expansion.

Recommendations

Because slot accounting systems cannot identify the origin of payouts,

estimating or measuring the impact of free-play oers is critical in crafting an eective

tax treatment policy. This policy decision seems to rest on the evidence of tax base

expansion. The following paragraphs describe how existing academic models can help

examine the extent to which the tax base is expanded, if at all.

Table 2 illustrates the possible outcomes and potential policy correlates.

Notice that no scenario would result in the complete elimination of a tax deduction, as

redeemers are at least required to wager the free-play awards once before leaving the

casino with any surviving balance.

Table 2

Potential Free-play Redemption Outcomes and Tax Policy Correlates

Redemption Outcome Policy Correlate

1. FP credits are lost and play is terminated. 1. Allow deduction at face value of

FP redeemed.

2. FP credits are partially lost and player walks with the

surviving balance.

2. Limit FP deduction accordingly.

3. FP credits are lost and additional funds from the

player’s own bankroll are lost.

3. Allow deduction at face value of

FP redeemed.

All of the redemption outcomes referred to in Table 2 would be determined

from the results of daily coin-in models designed to measure the campaign-level impact

of free-play redemptions. Lucas and Spilde (2017) and Suh et al. (2014) oer models

well suited to this task. Specic attention should be paid to the following hypothesis:

DFP

> 1/WAFP, where B

DFP

is the regression coecient for the daily dollar value of

discretionary free-play redemptions and WAFP represents the weighted-average oor

par, or average house advantage. A failure to reject this hypothesis would support a full

recovery of redeemed free-play oers, indicating that a full face-value tax deduction

would be appropriate. If this hypothesis were rejected, then the magnitude of the

coecient would be used to determine the appropriate tax deduction.

From Table 2, the rst and third redemption outcomes describe scenarios in

which face-value deductions should be permitted. Failure to allow such deductions

would result in the taxation of phantom revenue, as previously described. Moreover,

the third redemption outcome indicates evidence of tax base expansion, so taxing

UNLV Gaming Research & Review Journal t Volume 21 Issue 1

authorities would not want to discourage these free-play campaigns. In the case of

the second redemption outcome, the model results would determine the extent of the

gaming tax deduction. For example, if the model results indicated that the operator

recovered only 80% of the free-play dollars redeemed, then the tax deduction would be

limited to 80% of the free-play dollars redeemed.

The recommendation for jointly conducted research may seem challenging,

but operators are basing their argument for preferential tax treatment on mere claims of

an expanded tax base. Notwithstanding the amount of empirical evidence that would

suggest otherwise, the onus of proof should fall on the operators. To date, the bulk of

academic ndings would suggest that taxpayers may be at least partially subsidizing

free-play oers. Legislators have an obligation to protect and manage the state’s tax

base. After all, the payouts emanating from free-play wagers do come from a shared till.

Moreover, the knowledge gained from the results of repeated analyses would

lead to more protable free-play programs, which would benet both operators and

taxing authorities. An outcome of increased operating prots and tax base expansion

would certainly represent a win-win scenario. The academic approaches discussed

herein are intended to align the interests of the regulators and the operators. A free-

play program that is not working benets neither side. Such failures result in decreased

prots for operators and GGR deductions with no tax base expansion for taxing

authorities. Only the recipients of the free-play rewards would benet from overly

generous campaigns.

Although it’s clearly not without its challenges, the benets of this approach

should far outweigh the costs. Ultimately, the data should play at least some role in

shaping the tax policy, especially with regard to tax breaks for operators. Mere claims

and assumptions cannot comprise the sole basis for important tax policy decisions.

Competitive Challenges

As is the general case in markets saturated with loyalty programs, no operator

wants to be the rst to terminate theirs, or even roll back benets (Cigliano, 2000;

Uncles et al., 2003). This is true for free-play oers as well, as they are the core of

casino loyalty programs (Lucas & Spilde, 2017; Klebanow, n.d.). The same condition

holds for free-play tax policy at the jurisdictional level. For example, on the east coast

of the U.S. many jurisdictions exist within close proximity of one another. Many have

argued that restrictive free-play tax policies could impair operators from competing

against adjacent and less restricted jurisdictions (Armon, 2015; Belko, 2016; Klebanow,

n.d.; Murphy, 2016). Philander (2013) makes a similar argument with regard to

disparate ad valorem gaming tax rates, as applied to casinos operating on opposite sides

of a jurisdictional border.

At the regional level, free-play tax policies could evolve into something of

a prisoner’s dilemma. Specically, legislators may feel pressure to oer policies that

match those in neighboring jurisdictions to prevent customers from traveling across

jurisdictional boundaries. Academic research can be particularly helpful in such

conditions. For example, studies can estimate dicult-to-measure impacts associated

with free-play campaigns, rather than discussing the issue solely in terms of alleged

abilities of internal resorts to compete against other jurisdictions.

Free-Play Tax Deduction

UNLV Gaming Research & Review Journal t Volume 21 Issue 1

Countries such as Australia, China, and South Korea have limited the eect

of these border issues by instituting omnibus constraints on external gaming-related

oers such as discretionary free-play. While such policies serve to limit the swelling of

cost curves, they are more dicult to institute in less cooperative spaces. For example,

in the United States, multiple adjacent jurisdictions reside within a common national

space. A similar issue exists for European operators, given the close proximity of their

jurisdictional boundaries.

Future Research

A longitudinal study related to the ability of free-play campaigns to generate

a greater number of trips would also provide critical insight into the debate on tax base

expansion. Spectrum (2014) references such a study by New York regulators, but the

description revealed serious design problems related to self selection bias and wholesale

omission of key covariates. Any future study would need to randomly assign subjects

to control and experiment groups and track their behavior over time. Even then, there

would be design challenges related to the exposure of the control group to free-play

oers from competitors and cross-talk among study participants. Any sponsor of such

a study would have to be committed to the cause, as collateral damage to customer

relations is possible.

This issue could be further investigated by way of a more formal tax theory

paper. Such a paper would include a framework for evaluating a specied set of tax

policy outcomes. These outcomes would be evaluated from perspectives such as social

welfare, the casinos, and the patrons of the casinos. This would provide a broader

examination of any potential tax policy implications.

Acknowledgement: This research was made possible through funding from the Caesars

Foundation.

UNLV Gaming Research & Review Journal t Volume 21 Issue 1

References

American Gaming Association (AGA). (2015). By the Book Database, Taxation and

Revenue Sharing. Retrieved from http://www.gettoknowgaming.org/by-the-book

Armon, R. (2015, April 13). Ohio lawmaker wants casinos, racinos to pay taxes on

promotional credits. Beacon Journal. Retrieved from http://www.ohio.com/

news/local/ohio-lawmaker-wants-casinos-racinos-to-pay-taxes-on-promotional-

credits-1.583121

Barker, J. (2015, April 24). In varying amounts, casinos dole out millions on ‘free play’.

The Baltimore Sun. Retrieved from http://touch.baltimoresun.com/#section/-1/

article/p2p-83382228/

Belko, M. (2016, April 5). Pennsylvania casinos could face new tax under governor’s

proposal. Pittsburgh Post-Gazette. Retrieved from http://www.post-gazette.com/

business/development/2016/04/05/Pennsylvania-casinos-could-face-new-tax-

under-governor-s-proposal/stories/201604050061

Binkley, C. (2001, September 7). Reversal of fortune. Wall Street Journal, A1, A8.

Burns, R. (2010). Revenue sharing and free play. Indian Gaming, 20(4), 56-57.

Brunt, C.D. (2015, June 17). Racino ‘free-play’ tax issue left up to courts. Albuquerque

Journal. Retrieved from http://www.abqjournal.com/599739/racino-free-play-

tax-issue-left-up-to-courts.html

Cantrell, J. (n.d.). Freeplay accounting. Retrieved from http://gamingandleisuremagazine.

com/gl_articles/freeplay-accounting/

Casino Arizona. (2017). Retrieved from https://webcache.googleusercontent.

com/search?q=cache:5-lGSub3lX0J:https://www.casinoarizona.com/

promotions/500k-guaranteed-cash-giveaway/+&cd=1&hl=en&ct=clnk&gl=us

Cigliano, J., Georgiadis, M., Pleasance, D. & Whalley, S. (2000). The price of loyalty.

McKinsey Quarterly, No. 4, 68-77.

Fine, R. A. (2009). Free play isn’t free. Casino Enterprise Management, June, 60-61.

The Analyst, (2016, July 26). Gaming Today. Vegas could hit surprising baccarat record.

Retrieved from https://www.gamingtoday.com/industry/article/61719-Vegas_

could_hit_surprising_baccarat_record

Gruetze, M. (2012, April 12). Slot fans, casinos in midst of ‘golden age of free

play’. TribLIVE. Retrieved from http://triblive.com/x/pittsburghtrib/ae/

gambling/s_790232.html#axzz3h8gj4BFZ

Gu, Z. (2003). Casino marketing and promotion: Raising the eciency. Casino

Enterprise Management, 1(11), 46-47.

Inside Asian Gaming (2011, January 22). VIP credit risk, a micro, not so much macro,

issue in Macau. Retrieved from http://www.asgam.com/index.php/news/

item/1020-vip-credit-risk-a-micro-not-so-much-macro-issue-in-macau

Free-Play Tax Deduction

UNLV Gaming Research & Review Journal t Volume 21 Issue 1

Kilby, J., Fox, J. & Lucas, A.F. (2004). Casino Operations Management, 2

ed.

Hoboken, NJ: Wiley.

Kollars, B. & Otte, J. (2015, March 31). Gambling bill calls for taxing ‘free play.’

Dayton Daily News. Retrieved from http://www.daytondailynews.com/news/

news/gambling-bill-calls-for-taxing-free-play/nkjhQ/

Klebanow, A. (n.d.). Emerging issues in the use of free play. Urbino.net. http://www.

urbino.net/articles.cfm?specicArticle=Emerging%20Issues%20in%20the%20

Use%20of%20Free%20Play

Lucas, A.F. (2004). Estimating the impact of match-play promotional oers on the

blackjack business volume of a Las Vegas hotel casino. Journal of Travel &

Tourism Marketing, 17(4), 23-33.

Lucas, A.F. & Bowen, J.T. (2002). Measuring the eectiveness of casino promotions.

International Journal of Hospitality Management, 21(2), 189-202.

Lucas, A.F., Dunn, W.T. & Singh, A.K. (2005). Estimating the short-term eect of free-

play oers in a Las Vegas hotel casino. Journal of Travel & Tourism Marketing,

18(2), 53-68.

Lucas, A.F. & Kilby, J. (2012). Introduction to Casino Management. San Diego:

Gamma.

Lucas, A.F. & Kilby, J. (2008). Principles of Casino Marketing. San Diego: Gamma.

Lucas, A.F., Kilby, J. & Santos, J. (2002). Assessing the protability of the premium

player segment. Cornell Hotel & Restaurant Administration Quarterly, 43(4),

65-78.

Lucas, A.F. & Spilde, K.A. (2017). Estimating the eect of casino loyalty program

oers on slot machine play. Cornell Hospitality Quarterly. Advance online

publication. doi: 10.1177/1938965516686113

Murphy, S. (2016, April 4). Plainridge’s recent revenue surge may be a

mirage. Boston Globe. Retrieved from https://www.bostonglobe.com/

metro/2016/04/03/plainridge-boosts-revenue-with-generous-oers-free-play/

PimOmAO9YoX2UCfTVoNtoM/story.html

Nevada Revised Statutes (1997a), Chapter 463, Section 0161, Part 1(a-c). Retrieved

from https://leg.state.nv.us/nrs/NRS-463.html#NRS463Sec0161

Nevada Revised Statutes (1997b), Chapter 463, Section 0161, Part 2(c, g). Retrieved

from https://leg.state.nv.us/nrs/NRS-463.html#NRS463Sec0161

Philander, K. (2013). A normative analysis of gambling tax policy. UNLV Gaming

Research & Review Journal, 17(2), 17-26.

Pollack, M. (2010, November). Casino Tax policy: Identifying the issues that will

determine the optimal rate. Paper presented at the 103

Annual Conference on

Taxation, Chicago, IL. Retrieved from http://www.spectrumgaming.com/dl/

spectrumnationaltaxassociation.pdf

UNLV Gaming Research & Review Journal t Volume 21 Issue 1

Robison, J. (2014, January 29). Ask the slot expert: Should I play from a ticket or cash?

Casino City Times. Retrieved from http://robison.casinocitytimes.com/article/

ask-the-slot-expert-should-i-play-from-a-ticket-or-cash-62923

Rüdisser, M., Flepp, R. & Franck, E. (2015). Do casinos pay their customers to become

risk averse? Revising the house money eect in a natural experiment. Working

paper, Department of Business Administration, University of Zurich, Zurich,

Switzerland. Retrieved from http://www.business.uzh.ch/en/forschung/wps.html.

SCA Promotions. (2017). Luxor gives away more than $500,000 in big money basketball

promotion. Retrieved from http://www.scapromotions.com/press-releases/luxor-

gives-away-500000-big-money-basketball-promotion/

Seminole Hard Rock. (2017). April fool’s day sweepstakes and beat the taxman

sweepstakes. Retrieved from https://www.seminolehardrocktampa.com/casino-

events.htm

Sortel, N. (2010). “Casinos balance free play as part of their marketing strategy,”

SunSentinel, 12 November, available at http://articles.sun-sentinel.com/2010-11-12/

business/-casino-free-play-20101112_1_american-casino-guide-magic-city-

casino-steve-bourie

Spectrum Gaming Group (2014, August 18). Responding to competition, new challenges:

Regulatory, tax reform in Indiana. Linwood, New Jersey: Spectrum Gaming

Group.

Suh, E. (2012). Estimating the impact of free-play coupon value on players’ slot gaming

volumes. Cornell Hospitality Quarterly, 52(2), 134-143.

Suh, E., Dang, T. & Alhaery, M. (2014). Examining the eects of various promotion types

on slot gaming revenues. International Journal of Hospitality Management, 37,

68-79.

Stations Casinos. (2017). Pedal to the money. Retrieved from https://www.sclv.com/

Promotions/Gaming-Promotions.aspx

Thaler, R.H. & Johnson, E.J. (1990). Gambling with the house money and trying to break

even: The eects of prior outcomes on risky choice. Management Science,

36(6), 643-660.

Thompson, W.N. (2015). Gambling in America (2

ed.). Santa Barbara, CA: ABC-CLIO,

74.

Tversky, A. & Kahneman, D. (1991). Loss aversion in riskless choice: A reference-

dependent model. The Quarterly Journal of Economics, 106(4), 1,039-1,061.

Tversky, A. & Kahneman, D. (1992). Advances in prospect theory: Cumulative

representation of uncertainty. Journal of Risk & Uncertainty, 5(4), 297-323.

Uncles, M.D., Dowling, G.R. & Hammond, K. (2003). Customer loyalty and customer

loyalty programs. Journal of Consumer Marketing, 20(4), 294-316.

Free-Play Tax Deduction

UNLV Gaming Research & Review Journal t Volume 21 Issue 1