The demand for slot machine and pari-mutuel horse race wagering at a racetrack-casino

Abstract

This article expands the limited literature on the demand for casino (racino) wagering. The focus of this article is on the interrelationship of the demands for slot machine and horse race wagering at a racino. Wagering demand models were estimated for the two racino products; slot machine gaming and pari-mutuel horse racing. Horse race and slot machine wagering both decreased when a competing casino entered the racino's market area. As in earlier studies, slot machine customers did not, on net, bet on horse racing but horse racing customers did bet on slot machines. Pari-mutuel horse race wagering fell 21% after slot machines were introduced and 16% following the introduction of table games. On the other hand, slot machine wagering increased 13% in the presence of live horse racing and 14% in the presence of import simulcast horse racing. Slot machine wagering fell 8% with the introduction of table games. This has important policy implications if stakeholder shares of table game revenue are different than their corresponding slot machines shares. Those with lower table game shares may lose net gaming revenue if table games do not produce enough revenue to offset the expected loss of slot machine revenue.

Notes

¹ Association of Racing Commissioners, International, Inc., Pari-mutuel Racing, A Statistical Summary. Handle is adjusted for inflation using the CPI-U (1982 to 1984 = 100), US Department of Labor, Bureau of Labor Statistics.

² With the exception of pari-mutuel wagering in states where it is permitted, internet wagering on other forms of gaming is illegal in the United States under federal law. Internet and telephone wagering are conducted through a system called Advance Deposit Wagering (ADW).

³ For a discussion of the effects of competition on horse race wagering see Thalheimer and Ali (2008a).

⁴ Slot machine VLT gaming was permitted in 1990 at Mountaineer Park a pari-mutuel horse racetrack in West Virginia on an experimental basis under the auspices of the West Virginia State Lottery.

⁵ The 15 states which permit state-regulated slot machine racino gaming are: Arkansas, Delaware, Florida, Indiana, Iowa, Kansas, Louisiana, Maine, Maryland, New Mexico, Oklahoma, New York, Pennsylvania, Rhode Island and West Virginia. Racinos are permitted in Kansas but no license has been issued to-date. Minnesota permits card rooms but not slot machine gaming at its two racetracks.

⁶ See McQueen (2009).

⁷ See McQueen (2009).

⁸ American Gaming Association (2009).

⁹ Angst (2005).

¹⁰ Iowa Racing and Gaming Commission (2004) annual report.

¹¹ There are a number of studies which have examined the determinants of casino revenue but these are not demand studies since the dependent variable (revenue) is the demand variable (wagering) multiplied by its own price (win per cent) one of its determinants. Examples of these studies are: Cargill and Eadington (1978), Nichols (1998a, b), Hunsaker (2001), Levitzky et al. (2000) and Thalheimer and Ali (2008b).

¹² The demands for soccer and numbers betting are also examined in this article.

¹³ In another study of betting in the UK, Paton et al. (2004) examined the demand for bookmaker betting aggregated over all types of wagering offered at betting shops including horse and dog race betting, sports betting and slot machine (fixed-odds betting terminal) betting.

¹⁴ US Department of Commerce, Bureau of Economic Analysis.

¹⁵ http://www.iowa.gov/irgc/

¹⁶ Iowa Racing and Gaming Commission, Annual Reports.

¹⁷ Iowa Racing and Gaming Commission (1993) annual report.

¹⁸ See Iowa Racing and Gaming Commission web site: ‘Chronology of the Iowa Racing and Gaming Commission’, http://www.iowa.gov/irgc/Chronology.htm. See Iowa Racing and Gaming Commission (2004) annual report for table games launch date.

¹⁹ Iowa Racing and Gaming Commission Annual Reports.

²⁰ Import simulcast wagering also included greyhound races. Live greyhound racing was not conducted at the racetrack. The dependent variable did not include import greyhound handle, 12% of on-track wagering, to allow for a more direct connection between on-track horse race wagering and its determinants.

²¹ The CPI-U (1982 to 1984 = 100), US Department of Labor, Bureau of Labor Statistics, was used to compute the real dollar values of all monetary variables.

²² Thalheimer and Ali (2003) found that the demand for slot machine wagering was negative at lower levels of per-capita income and positive at higher income levels.

²³ A racetrack programme includes a full complement of races at a racetrack over a given day.

²⁴ Import simulcasts figures disaggregated by race horse breed were not available.

²⁵ See for example, Ali and Thalheimer (1997, 2002), Thalheimer (1998, 2008) and DeGennaro (2009).

²⁶ Festival of Racing source: Prairie Meadows Racetrack and Casino (1993–2006). Quarter Horse racing source: Iowa Quarter Horse Racing Association.

²⁷ Iowa Racing and Gaming Commission, 2000 Annual Report.

²⁸ Ibid.

²⁹ LM is distributed as χ ²(p), where p is the number of lagged residuals.

³⁰ LM is distributed as χ ²(z), where z is the number of regressors.

³¹ To determine long-run effects of individual variables in the slot machine demand model, the coefficient of interest in Equation 1 is divided by (1 − α ₁), where α ₁ is the coefficient of the stochastic trend variable, ln(RSLOTHAND(−1), in Equation 1 with corresponding value in Table 1.

³² % impact of table games = [exp{(α14/(1 − α1))(57 − 0)} − 1]100, where α ₁ and α ₁₄, are the coefficients of ln(RSLOTHAND(−1)) and TABLES, in Equation 1 with corresponding values in Table 1.

³³ ε _SLOTWINPCT = {α12/(1 − α1)}SLOTWINPCT.

³⁴ SLOTWINPCT_max = 1/{α12/(1 − α1)}.

³⁵ % impact of riverboat in market area = [exp{(α22/(1 − α1))(1 − 0.07)} − 1]100, where α ₁ and α ₂₂, are the coefficients of ln(RSLOTHAND(−1)) and BOATWINPCT, in Equation 1 with corresponding values in Table 1.

³⁶ % impact of temporary closure of Tama Indian casino = [exp{(α21/(1 − α1))(1 − 0)} − 1]100, where α ₁ and α ₂₁, are the coefficients of ln(RSLOTHAND(−1)) and INDIANCLOSE, in Equation 1 with corresponding values in Table 1.

³⁷ % impact of live gallop horse race days = [exp{(α15/(1 − α1))(20− 0)} − 1]100, where α ₁ and α ₁₅, are the coefficients of ln(RSLOTHAND(−1)) and DAYSGALLOP, in Equation 1 with corresponding values in Table 1.

³⁸ % impact of import simulcast horse race programmes = exp{(α17/(1 − α1))(15 − 4)} + (α ₁₈/(1 − α ₁))(15² − 4²)} − 1, where α ₁ is the coefficient of ln(RSLOTHAND(−1)) in Equation 1 and α ₁₇ and α ₁₈ are the coefficients of SIMHORSE_DAY and SIMHORSE_DAY² in Equation 1 with corresponding values in Table 1.

³⁹ The QUALITYRACES effect also reflects a JULY effect since the two variables were highly correlated. % impact of high quality races = [exp{(α15/(1 − α1))(1 − 0)} − 1]100, where α ₁ and α ₂₀, are the coefficients of ln(RSLOTHAND(−1)) and QUALITYRACES, in Equation 1 with corresponding values in Table 1.

⁴⁰ To determine long-run effects of individual variables in the slot machine demand model, the coefficient of interest in Equation 2 is divided by (1 − β ₁), where β ₁ is the coefficient of the stochastic trend variable, ln(RPARIHAND(−1)), in Equation 2 with corresponding value in Table 2.

⁴¹ % impact of one live gallop horse day = [exp{(β14/(1 − β1))(1 − 0)} + (β ₁₅/(1 − β ₁))(1² − 0²)} − 1]100, where β ₁ is the coefficient of ln(RPARIHAND(−1)) in Equation 2 and β ₁₄ and β ₁₅ are the coefficients of DAYSGALLOP and DAYSGALLOP² in Equation 2 with corresponding values in Table 2. % impact of one harness horse race day = [exp{(β16/(1 − β1))(1 − 0)} − 1]100, where β ₁ and β ₁₆, are the coefficients of ln(RPARIHAND(−1)) and DAYSHARNESS, in Equation 2 with corresponding values in Table 2.

⁴² % impact of one import simulcast racetrack programme per race day = [exp{(β17/(1 − β1))(1 − 0)} − 1]100, where β ₁ and β ₁₇, are the coefficients of ln(RPARIHAND(−1)) and SIMHORSE_DAY, in Equation 2 with corresponding values in Table 2.

⁴³ The QUALITYRACES effect also reflects a JULY effect since the two variables were highly correlated. % impact of high quality races = [exp{(β19/(1 − β1))(1 − 0)} − 1]100, where β ₁ and β ₁₉ are the coefficients of ln(RPARIHAND(−1)) and QUALITYRACES, in Equation 2 with corresponding values in Table 2.

⁴⁴ % impact of riverboat in market area = [exp{(β21/(1 − β1))(1 − 0.07)} − 1]100, where β ₁ and β ₂₁, are the coefficients of ln(RPARIHAND(−1)) and BOATWINPCT, in Equation 2 with corresponding values in Table 2.

⁴⁵ % impact of slot machines = [exp{(β12/(1 − β1))(1600 − 0)} − 1]100, where β ₁ and β ₁₂, are the coefficients of ln(RPARIHAND(−1)) and SLOTS, in Equation 2 with corresponding values in Table 2.

⁴⁶ % impact of table games = [exp{(β13/(1 − β1))(57 − 0)} − 1]100, where β ₁ and β ₁₃, are the coefficients of ln(RPARIHAND(−1)) and TABLES, in Equation 2 with corresponding values in Table 2.

⁴⁷ % combined impact slot machines and table games = {(1 + gslots)(1 + gtables) − 1}100, where g = growth rate, g _slots = −0.22 and g _tables = −0.16.

⁴⁸ This is the same practice as that for all state-regulated casinos in the United States where no distinction is made between statutory distributions to stakeholders from slot machine or table game revenues.