MEMORANDUM

INTRODUCTION:

EGDs (Electronic Gaming Devices) or VGDs (Video Gaming Devices) are probably the most widely adopted innovations in contemporary casino gaming. While these devices are essentially equivalent to slot machines, they offer the added attractions of programmed games such as keno, blackjack, and several variants of poker. The impact on revenues of these high-tech slot machines is quite significant: roughly 60% of gross gambling revenues in Las Vegas and New Jersey casinos are attributed to these machines.[1] Consequently, their revenue generating potential was immediately recognized by several state and local governments, which have decided to legalize them. Video Lottery Terminals or VLTs are the same as regular video gaming devices but are referred to as “lottery terminals” because they are regulated by the state’s lottery commission, which receives a share of the revenues.

VLT gaming was first legalized in South Dakota in 1989. Since then West Virginia (1990), Oregon (1992), Rhode Island (1992) and Delaware (1995) have followed suit. Some states, like Delaware and Rhode Island, have authorized the operation of VLTs in close proximity to pari-mutuel wagering sites such as racetracks in order to enable these facilities to compete with the booming casino market of the 90s. The experience of these states in terms of generating more revenues or resuscitating their ailing racetracks[2] has been overwhelmingly positive. Thalheimer (1998) presents empirical evidence from the experience of Mountaineer Racetrack in West Virginia that adding VLTs into the gambling portfolio mix negatively impacts pari-mutuel revenues from horse wagering but more than compensates for this through larger revenue gains from the VLTs.[3] Hence, one proposal to help shore up the educational funding deficit in New Hampshire is to legalize the operation of VLTs or VGDs (Video Gambling Devices) in four racetracks and two hotels located within the state. Licensing and tax revenues from the operation of these machines can then be earmarked specifically for state educational expenditures. The proposed locations for these devices include The Balsams Hotel (Dixville Notch), Mount Washington Hotel (Bretton Woods), Hinsdale Greyhound Park (Hinsdale), Lakes Region Greyhound Park (Belmont), Rockingham Park (Salem), and Seabrook Greyhound Park (Seabrook).

This memo examines the revenue potential of VLTs with due consideration to its cannibalization[4] effects, particularly on pari-mutuel wagering (i.e. wagering on horse and dog races) and state lottery revenues. Furthermore, this memo provides forecasts of the social costs based on national studies conducted on problem gambling and on the experience of other states that have legalized VLTs. Net revenue estimates for the state of New Hampshire can then be examined.

On the revenue side, two approaches are used in attempting firs-pass approximations. The first approach adopts the common practice of using the average of net revenue numbers for slot machines located in other states. By simply multiplying these net revenue statistics by the intended number of machines in New Hampshire, a very rough approximation of the revenue potential can be made. However, this approach can be criticized on several grounds. First, one can make the argument that the net revenue per machine in New Hampshire may not be comparable to those in other states due to the differences in market demographics. Second, the net revenue per machine may be sensitive to the number of machines in place; too many machines may end-up lowering the net revenue of each individual machine. Conversely, too few machines may result in a lower incentive for VLT players to make the trip, because of concerns about machine availability.

The second approach evaluates the New Hampshire gambling market.  A 125-mile radius is used to determine the markets for each of the proposed VLT sites in New Hampshire. After correcting for overlaps with casinos and other racetracks with VLTs, a unified market base is identified in order to project total state revenues for all of the facilities. To implement these projections, calculations on per capita gambling by the National Opinion Research Center (NORC) at the University of Chicago are utilized.

With respect to social costs, baseline estimates are taken from the Gambling Impact and Behavior Study (1999)[5] conducted by the NORC. Per capita prevalence rates of problem and pathological gambling as well as costs per gambler are used to project total annual social costs for all 6 proposed VLT sites in New Hampshire.

REVENUE ESTIMATES:

                 One way to provide a rough estimate of the revenue stream from installing VLTs in New Hampshire is to simply take an average net revenue number for a machine currently in operation in some other state and multiply this by the proposed number of machines to be authorized. For this method to be reliable, several assumptions need to be made. First, the market size and demographics between the two sites need to be similar. Second, the number of machines in both markets needs to be roughly equal. The reason for this is that slight subtleties in market scenarios are just as conceivable: 100 VLTs in a small market and 1000 VLTs in a larger market may give out the same annual average net revenue per machine, ceteris paribus. These scenarios clearly imply that many factors need to be similar (or many demand and supply conditions need to be just so) in order for this method to be accurate. This notwithstanding, the range of outcomes can still be of some use, if only as a first-pass approximation of the revenue numbers one can expect. Table 1 below shows the range of potential revenue outcomes assuming a total of 5000 machines.[6]    

Table 1: Projected Revenues Based on Net Revenues per Machine in Other States, 1999

*Assuming 52 weeks of operation per year.

**Assuming 44.8% State percent tax.

Source: La Fleur’s Fiscal 1999 VLT Special Report. Teresa La Fleur, Editor. TLF Publications. Boyds, MD. 1999.

                Given the wide range of revenue outcomes, a specific net-revenue-per-machine number from a market similar to New Hampshire’s must be selected. From a general analysis of population and median income statistics, the Delaware market seems to be the most similar to that of New Hampshire.

In determining the market areas and population sizes of the Delaware and New Hampshire gambling markets, all casinos and pari-mutuel facilities with VLTs or slot machines in the general area are identified. A circle with a 125-mile radius[7] is then drawn around each of these sites. The specific criteria for identifying the relevant casinos and pari-mutuel facilities is that all potential competitors whose market areas overlap those of the proposed VLT sites in New Hampshire need to be included in the analysis. These gambling sites are presented in Table 2.

The market area for each of the VLT sites can then be identified, by using perpendicular bisectors to apportion the market based purely on distance.[8] The areas controlled by each VLT site in New Hampshire can then be added to show the total market area that the state’s gambling facilities can generate revenues from. The same procedure is used to identify the aggregate gambling market in Delaware, which has 3 VLT sites: Delaware Park Racetrack and Slots (Wilmington), Dover Downs Slots (Dover), and Midway Slots and Simulcast (Harrington).  Finally, in order to evaluate the proposition that the Delaware and New Hampshire gambling markets are sufficiently comparable (i.e. so that net revenues per machine in Delaware can be used to project revenues in New Hampshire), the population size and the income demographics of these two markets must (at the very least) be relatively similar.  

Table 2: Selected Existing and Proposed Gambling Facilities

New Hampshire Market*

Delaware Market*

*Data gathered from Internet search and phone interviews in July , 2000.

                There are some obvious similarities between the two markets. Both the New Hampshire and Delaware sites face considerable competition from neighboring facilities: over 14,000 slot machines in competing gambling sites for New Hampshire and well over 17,000 machines (mostly in Atlantic City) for Delaware. Gambling facilities in both states also face strong competition from popular tourist destination sites that offer gambling as only one of many other activities: New Hampshire sites will face competition from Foxwoods and Mohegan Sun to the South and Casino de Montreal to the North, while Delaware sites face significant competition from Atlantic City casinos.

With respect to market size, the population base or market size for New Hampshire is roughly 9 million with a median income range among the covered counties of $24,893 to $51,855.[9] The Delaware market has a larger population base of approximately 13 million with a median income range of about $19,412 to $66,745. Based on surveys conducted by NORC (1999) and the Wisconsin Policy Research Institute (1995),[10] the majority of casino patrons fall within the $20k to $50k income range, making both New Hampshire and Delaware prime gambling sites. The decomposition of the population numbers by county can be found in Tables 3A and 3B below.

Table 3: Population Estimates/Projections

A. The New Hampshire Market*

B. The Delaware Market*

*1999 growth rates used to calculate projections for 2000 and 2001, and market shares in percentages are indicated after each state.

**Projections for 2004 from Rand McNally Commercial Atlas and Marketing Guide, 2000. 

Source:http://www.census.gov/datamap/www/index.html

While these population numbers seem to indicate that Delaware VLT sites have a significantly larger market than that of the proposed New Hampshire sites, it is important to note one cogent argument that may lead us to conclude otherwise. In determining the New Hampshire and Delaware markets, only the influence of distance has been considered so far. While this is obviously of paramount concern for a market predominantly composed of day-trippers, there is also the gravity effect from being proximal to a popularly known tourist spot (i.e. Atlantic City or Foxwoods).

It follows that the population base for both the Delaware and New Hampshire sites need to be scaled down in order to account for this effect. Incidentally, it is important to note that this argument applies more so for the case of Delaware Park in Wilmington, DE which is much closer to Atlantic City, NJ (i.e. approximately 65 miles away) than the closest New Hampshire sites are to Foxwoods (i.e. from Rockingham Park in Salem, NH, 99 miles), Mohegan Sun (i.e. from Rockingham Park, 100 miles), or even Casino de Montreal (i.e. from The Balsams, 120 miles). Moreover, Foxwoods may be the largest casino in the world with over 4500 slot machines, but it pails in comparison to the combined gambling supply of Atlantic City. The gravity effect of Foxwoods on the proposed VLT sites in New Hampshire will be much lower than the effect of Atlantic City on the Delaware sites.

Therefore, there is every reason to expect that the Delaware market is much lower than 13 million, and that this estimated market size should be viewed as a very generous upper bound estimate. If one makes the not unreasonable assumption that all New Jersey residents would much prefer Atlantic City (15,000+ slots) over Delaware Park Racetrack and Slots (2000+ slots) in Wilmington, then the market for Delaware is closer to 10 million people (i.e. by subtracting the New Jersey residents from the Delaware market), and is not too far off from the 9 million estimate for the New Hampshire market.  Hence, a  case can indeed be made for using the net revenue numbers from Delaware machines in order to make projections of expected VLT revenues in New Hampshire. From earlier calculations presented in Table 1, New Hampshire stands to gain $242M in gross tax revenues from operating 5000 VLTs while imposing a 44.8% state tax on machine revenues.

                In order to test how robust these projected revenue numbers are, another method is used to forecast VLT revenues. The National Opinion Research Center (NORC) at the University of Chicago published a report in 1999 evaluating the impact of gambling trends in the United States. Studying 100 sample communities across the US from 1980 to 1997, the report produced baseline estimates of per capita casino spending. In order to impute expected VLT revenues from casino spending, use is made of the fact that slot machines account for around 60% of casino revenues in various gambling facilities in Atlantic City and Las Vegas.

However, slot machines in casinos will definitely pull in more revenues than slot machines located in racetracks or hotels. Gambling industry experts observe that video wagering machines in non-casino settings usually generate less revenues than those based in casinos, because the presence of table games in casinos “define the gaming atmosphere and stimulate the behavior within a [gambling] facility.”[11] Comparing the numbers for gross profits per machine in non-casino (i.e. Louisiana, Montana, Rhode Island) and casino sites (i.e. Colorado towns of Cripple Creek, Central City, and Black Hawk), 30% is identified as a more reasonable share number. The calculations are summarized in Table 4 below.

Table 4: Projected VLT Revenues from NORC Estimates*

*Based on NORC (1999) estimates of 1997 per capita casino spending.

**Population total within 50 miles of the VLT sites were drawn from all the counties of New Hampshire and Massachusetts (i.e. those falling within the market scope determined earlier), Oxford, Franklin and York  in Main, and Bennington, Windham, and Windsor in Vermont.

                With these additional calculations, projections of additional gross tax revenues for New Hampshire range from $200M to $240M.

COST ESTIMATES:

                To a large extent, estimating the economic and social costs of operating VLTs  or slot machines is a far more complicated exercise than projecting its revenues. In terms of economic costs, existing infrastructure will depreciate far more quickly due to increased traffic, and new facilities (i.e. parking space and access roads) will need to be constructed in order to maximize patronage. As regards social costs, we again rely on the NORC (1999) study, which estimated the costs attending pathological and problem gambling in the US.[12] These costs included expenses related to job loss, bankruptcy, arrests, corrections and poor health, to name a few. While these estimates do not account for the time dimension of problem and pathological gambling[13] the estimated numbers can still serve as a useful benchmark.