In our recent articles, we determined your marketing battleground areas and market penetration analysis. The goal of these layers of data and analysis is to build a model that results in what is commonly called a gravity model. In this article, we will go over the final processes necessary to get to this end goal.
The gravity model of migration is a model in urban geography derived from Newton’s law of gravity, and is used to predict the degree of interaction between two places.1 William J. Reilly was inspired by Newton’s law of gravity to create an application of the gravity model to measure retail trade between two cities. In 1931, he developed Reilly’s law of retail gravitation, which, in summary, says market centers will divide the market between themselves and, to simplify, customers will shop only at one center or the other.2
Reilly’s Law is relied upon heavily to build gravity models, which have become a constant in gaming’s expansion. In these models, travel is reasonably equidistant between Casino A and Casino B. The second factor is – and we feel this is the most important assumption – the customers are basically indifferent between the casinos. We have long held that in today’s world of “casinos in most every state,” this indifference can also be called the commoditization of casinos. For some who adhere to older and very outdated processes, commoditization is not a factor and they often use attractiveness. We find that attractiveness is largely archaic in the casino-commodity world and is best used only when looking at two properties within very close proximity to one another and that they have obvious tangible
physical property differences. Internally, when we build a gravity model we find this process to be as much science as art. As for the art, the analyst’s experience in operations and marketing of casinos, as well as their intuition of external factors, are important in determining the significance of measuring the science of distance and especially the effect of competition. The science process is related to using tested techniques from data sourcing to actual model building in a consistent manner. To be considered science, the final results must be reasonably duplicated and tested. “It’s a standard principle in science: If you can’t replicate a set of results, then there is a problem with it. A flaw or a fraud is at work. Either you made a mistake, or you made it up.”3 In that regard, the gravity modeling process we utilize is repeatable and can be tested for reasonability through other independent peer review.
Reviewing the needs list
Previously we discussed the basic needs to collect a reasonable set of data for a gravity model. These were geo-mapping distances that should reasonably be considered the market trading range. Secondly, quantifying the competition for counts of slot machines, table games, hotel rooms and amenities within, and more importantly, which are on major highways which could act as “interrupters” to the site in question.
The last hurdle is to assemble data at a zip code level on population, incomes and demographics and the gaming value of potential customers. This would be accomplished by using published federal and state government databases, universities, and commercial and trade agencies, which specialize in such databases and information. As an additional component, if you are building models to support more than just today, typically you will need to have a forecast for three to five years forward. To satisfy that requirement, your planning will need to research and document the growth rates for population, incomes and demographics and gaming propensity.
Putting the models together
We collect data from the various state sources including the American Gaming Association, National Indian Gaming Commission, US Census Data-Population Division and we constantly refresh to have most recent data available.
The first place to start is one of the easier to derive and that is: What does a person spend in a year on casino gaming? We take the national figures for gaming revenue and then divide that by the number in population who are equal to or more than 21 years of age to come up with an annual gaming budget. From the American Gaming Association research, 34% of adults visited a casino, while 32% say they actually gambled. When those are all considered, we estimate an average annual gross gaming budget on a per-person basis for those likely to gamble.
Once we have the gaming budget, we start to look at the geography within the trading range for the property. Populations living closer to casinos experience a higher expected adult gaming population participation rate. According to Harrah’s research issued in 2006,4 the largest casino feeder markets all had gaming participation rates as high as 34% while the national average gaming participation rate was 25%. Additionally, since 2006, we believe this number has changed with an increased bias as casinos have become more prevalent. Based on the national average adult casino gaming participation rates and our research, we project a fair range for the gaming participation rate for adults living near casinos is upward of 41% and should be reduced the farther adults live from a casino gaming property.
Competition & Propensity to Gamble
We then apply a regional premium to reflect the populations with proximity and exposure to existing casinos having a higher experience rate or percentage of casino gamers than the national average. Casino gaming is dependent upon access to a population that will use the service/product provided. As a result, our population study is further detailed into specific geographic areas. While all of our raw data is zip code specific, we generally start with some sort of regionalization of the trading range for the property, using city, county, MSA or CBSA and roll-up the zip code data as required. Lastly, while state is also used, depending on the geography, you will generally need to be more granular.
Visually the combination of population and number of slot machines as a proxy for combination is shown on the map to the right. The map shows where the populations are and their size along with clusters of competition. Using simple visual process of elimination one can see if populations would need to drive past the competition to arrive at the project casino.
Forecasting Market GGR
We now are able to build a forecast for total market GGR by using the adult gaming population participation rate and per capita income figures within varying distances of the property, adjusted for any locational tendencies. For regional casinos, we generally consider two hours as the largest market share with less and less as the distance increases. We generally use mileage radius bands but also have used driving times, and more often use both for specific instances within documents. While driving times are intuitively more valuable, we find that with all of the extrapolation involved by the algorithms needed to build driving times, the use of radius bands has lower volatility and is thereby more useful.
Once we determine our trading range and outlying areas we can then go about building our forecast of the adult population and then reduce to the participation rate with a propensity to game. Each market area has its own household income and per person household income rates which we use to derive an estimated gaming budget. Using the estimated gaming budgets with the associated total adult gaming population rates indicates a gross gaming revenue projection for the market. A hypothetical example is shown in the following table.
We back test our overall results with the average yearly gaming budget for the market area as a percentage above or below the national average. The difference may be reflective of numerous items, including but not limited to: lower or higher levels of household income in this region as compared to US, and trips to other jurisdictions beyond the trading range market area utilized in our model.
Additional proof of the trading range annual GGR forecast would include use of local or regional gaming data and using figures for Win per Unit Day (WUD) as proxies for revenue tests. Building a range of WUD provides a low-to-high array of annual GGR potential. By this evaluation, we can determine if our projection is likely or unlikely to be achieved by the market.
Forecasting Property GGR
When calculating our proprietary capture rate for GGR, many different variables are utilized to determine fair estimates. The variables include, but are not limited to, the function of competition, distance hurdles to the closest casino, asset quality, and other forms of legalized gaming, not the least of which is: Does the drive distance area have other competing gaming options? Also note that some of each gaming budget could be spent out of the trading range or around the country in Las Vegas or Atlantic City, for example.
Applying our capture rates based on the discussed variables results in a projected GGR for the property, which we would review by various breaks of geography. With these established, we would forecast total GGR along with a sensitivity range for the property in the following table.
With these pieces acquired, you have the foundation for your gravity model. Our goal is to build a model that provides a reasonable accounting for gaming revenue from the market, which would take into consideration a patron’s distance from the facility and the effect of competition on total gaming revenue from the market.
The gravity model needs to have a solid build-up, as well as projections that are both reasoned and reasonable to see by the end user. Although differing methodologies/modeling approaches are likely between any parties making these models, the end user must be able to read through, digest, and reconcile the results in their mind. The analyst is the ‘expert’ on this assignment and it is the analyst’s work product, so it must be able to stand by the end result.
1 Rodrigue, J.P., Comtois, C., Slack, B. (2009), “The Geography of Transport Systems”, Routledge.https://en.wikipedia.org/wiki/Gravity_ model_of_migration, Accessed 6/19/2015
2 Amal Datta, “Reilly’s law of retail gravitation”, https://www.scribd. com/doc/70608682/Reilly-s-law-of-retail-gravitation. Accessed 6/15/2015
3 Sally Jenkins, “Why Roger Goodell might be in tough spot on Tom Brady suspension”, Washington Post,http://www.washingtonpost.com/sports/redskins/as-brady-appeal-nearsroger– goodell-is-stuck-in-a-corner-of-his-owncreation/ 2015/06/17/a5fcbaa6-1456-11e5-89f3-61410da94eb1_story.html, Accessed 6/18/15
4 “Harrah’s Survey: Profile of the American Casino Gambler, 2006,” Harrah’s Entertainment.
Jay Sarno has 20+ years of experience in the Hospitality and Gaming Industry. Jay consults on casino marketing segmentation programs, software product development and technology solutions evaluations, selections and implementations. Jay has implemented over 20 data warehouse systems and currently also teaches courses in Hospitality Management for Richard Stockton College of NJ. Jay can be reached at JSA2002@comcast. net and welcomes your comments and questions.