The United States Gambling Research Institute was asked by the Missouri Council on Economic Integrity (John Loudon, Chairman) to produce an impartial review of the study, The Economic Impact of Gaming in Missouri, by Charles Leven and Don Phares, prepared for Civic Progress of St. Louis, MO, and released in April, 1998.

About the United States Gambling Research Institute (USGRI)

          The Institute is a national center to which public officials, community leaders, press, and private citizens can turn for critically needed objective information about legalized gambling. It is a non-profit (501(c)3) organization established primarily with funds from the John D. and Catherine T. MacArthur Foundation to:

  • Work with public and private groups to provide impartial nonpartisan evaluations of the social, economic, and political implications of gambling proposals in their communities.

  • Share information and experience of communities nationwide with regard to gambling.

  • Conduct original research and evaluate the work of other researchers.

  • Examine the impact of gambling on the political process by looking at campaign contributions, lobbying, and the role of gambling revenues on public budgets.

  • Contribute to and monitor local, state, and federal research efforts related to legalized gambling.

  • Provide the media with an independent information source that has no financial ties to the gambling industry.

  • Sponsor workshops and conferences for public officials and community leaders.

  • Maintain a World Wide Web site with information about the impacts of legalized gambling.

United States Gambling Research Institute
245 Main Street     Northampton, MA 01060
TEL: (413) 584-0855 / FAX: (413) 585-0688
E-mail:     Web Site:

Return to top of page

Biographical Information

          Robert Goodman is the Executive Director of the United States Gambling Research Institute. He is an urban planner, economic development consultant and teaches at Hampshire College in Amherst, MA. Mr. Goodman's book The Luck Business (Free Press, 1996), a study of contemporary gambling policy, grew out of his work directing a two-year research project funded by the Ford Foundation and the Aspen Institute. He has testified before Congress, the National Gambling Impact Study Commission, and state legislatures, and has spoken frequently about issues of planning and economic development in this country and abroad.

          Edward Feser is a regional economist and urban planner who specializes in studies of industrial productivity and regional growth, industrial location, and local and regional economic development policy. He teaches in the Department of City and Regional Planning at the University of North Carolina at Chapel Hill and is a Fellow at the UNC Center for Urban and Regional Studies. He is co-author with Professor Emil Malizia of Understanding Local Economic Development, forthcoming from Rutgers University Press.

Understanding the Economic Impact of Casinos in Missouri

A Review of the Leven-Phares Study
Prepared for the United States Gambling Research Institute
by Robert Goodman and Edward Feser

The United States Gambling Research Institute was asked by the Missouri Council on Economic Integrity to produce an impartial review of the study, The Economic Impact of Gaming in Missouri, by Charles Leven and Don Phares, prepared for Civic Progress of St. Louis, MO, and released in April, 1998. The goal of our analysis was to examine the validity of the methodology and data used in the Leven-Phares Study. We conclude that revised economic impact estimates are warranted to provide a more comprehensive assessment of both the benefits and costs of the casino industry to the Missouri economy. It is our hope that this review and our revised impact estimates will contribute to a fuller and more accurate understanding of these impacts.


Major Findings of the USGRI Review:

1. The Leven-Phares Study Addressed Important Objectives and Used an Acceptable Impact Analysis Model.
          USGRI agrees on the importance of the stated objectives of the Leven-Phares study: in their words "to take an objective look at the economic impact of the gaming industry on the Missouri economy," and to determine "the bottom line economic impact." As they state, and we concur, "With such questions addressed and answered empirically, future debate can at least have some firm basis."
          The study was carried out with an approximated Missouri input-output model (called IMPLAN), with secondary data provided by the company that created the model and primary data from surveys of casino patrons and operators. The study's results are based on the total direct and multiplier impact of spending by the casino operations, net of the 'displacement effect' or consumer spending that would probably have occurred in other Missouri sectors in the absence of the casinos. Expenditures by the casinos include payments to employees (who spend some fraction of their wages in Missouri), purchases from Missouri vendors, and payments to state and local governments (in the form of taxes and fees).

2. The Leven-Phares Study Makes a Valid Attempt to Use Innovative Methods to Estimate Certain Economic Benefit Impacts.
          The Leven-Phares study is among few to date that attempt to accurately document displacement spending associated with casino gambling. Most studies adopt an 'either-or' approach, in which all spending made by casino patrons either is viewed as substitution for other spending (thus yielding no positive effect) or as completely new spending (thus generating a very large positive effect). The Leven-Phares analysis attempts to find a middle ground, where some spending by Missouri gamblers is, in fact, 'new' to the Missouri economy.

3. However, the Study Completely Omits the Economic Costs of Problem Gambling. USGRI Found No Valid Reason for the Decision to Omit These Costs.
          Leven-Phares themselves explicitly acknowledge that they have decided not to take into account certain factors relevant to the economic impact of casino gambling. In their words, "there are some economic or economic-related issues which are beyond the scope of this analysis." One such excluded factor was the cost of problem gambling to the state's economy. The failure to include any costs for problem gambling in Missouri seriously affects the validity of the study. Without full consideration of costs as well as benefits, an economic impact study cannot be a comprehensive basis for understanding the full impact of casino gambling on the Missouri economy.
          The study disregards extensive research elsewhere in the country which indicates a strong connection between increases in the availability of casino gambling and increases in problem gambling and the costs of problem gambling. Such existing research includes studies in other states, as well as a national survey of problem gambling prepared for the casino industry by researchers at the Harvard Medical School.

4. A Conservative Estimate for the State-Wide Costs Resulting from Increased Problem Gambling Due to the Casinos in 1996 Would Be At Least $182 Million. A Figure As High As $728 Million Would Be Plausible Based on Comparative Data.
          Problem gamblers create economic costs for both the public and private sectors. The costs are related to increased bankruptcies, unpaid debts, check fraud, embezzlement and other forms of economic theft engaged in by problem gamblers. The prosecution and incarceration of those problem gamblers who commit crimes is also costly. The consensus of existing national and state-wide research is that legalizing opportunities for casino gambling increases the prevalence of severe problem gamblers by at least 0.45 percent and possibly as high as 1.8 percent. Based on recent empirical studies of problem gambling, each problem casino gambler adds approximately $10,113 in yearly costs to a state's economy. Based on both those figures, we estimate that the losses to Missouri caused by increased problem gambling in 1996 range from a minimum of $182 million to a conservatively estimated high of $728 million.

5. There Appears to be a Bias Towards Overestimating Benefits in the Leven-Phares Study.
The following benefit categories may be overestimated:

          a) Impacts from Payments to Missouri Workers
          The Leven-Phares study estimated that 95 percent of the labor cost spending by the casinos goes to Missouri workers. However, since almost all of the casinos are located within easy commuting distance of other states, it is highly probable that significantly more than 5 percent of casino workers reside in neighboring states. If the percentage of Missouri casino workers is indeed lower than 95 percent, then total direct labor impacts (estimated at $200.6 million in 1996) and the additional rounds of induced spending that derive from wages are overestimated since workers are likely to spend much of their wages and salaries in their home communities. Since the Leven-Phares estimate is based on self-reported data by the casinos, and since casinos have an incentive to demonstrate Missouri benefits by overestimating the share of their workers residing in Missouri, the original data itself is questionable.

          b) Impacts from Casino Expenditures for Supplies and Equipment in               Missouri
Leven-Phares estimated that 92 percent of casino spending for supplies and equipment for 1996 ($237 million) was made within Missouri. The source of this percentage is again self-reporting by the casinos. Two factors make this high percentage doubtful. First, the fact that most casinos are located within metropolitan areas that span state boundaries suggests that the percentage of out-of-state spending is likely to be higher than 8 percent. Second, and perhaps more important, casinos that are subsidiaries of national firms are more likely to contract for a large percentage of their goods and services out-of-state. Most Missouri casinos are indeed owned by gambling companies headquartered outside of Missouri.

          c) Impacts of Dollars Recaptured by the Missouri Casinos
          Leven-Phares estimated how much money Missouri "recaptured" from gambling venues in other states by surveying Missouri casino-goers and asking them to estimate how much money they had spent on gambling and other activities in other states before casinos were available in Missouri. But since they surveyed only frequent Missouri visitors (those who had visited casinos three or more times in 1996) and then used this data to represent the average of all Missouri casino users, the recaptured dollars were likely overestimated. The 823 respondents to the Leven-Phares survey reported that they visited Missouri casinos an average of seventeen times in 1996. Such frequent gamblers are much more likely to report having traveled to out-of-state gambling venues when casinos were unavailable in Missouri. Extrapolating the findings for frequent gamblers to all Missouri gamblers overestimates the volume of total dollars recaptured by the casinos.

          d) Impacts of Dollars Captured by the Missouri Casinos
          In estimating the "displacement effect," Leven-Phares assume all money spent by out-of-state gamblers ($177 million) represents new spending in Missouri that would not take place without casino gambling. But some of this spending might have occurred even without casinos if some share of out-of-state gamblers visited Missouri primarily for other reasons (to visit family or other attractions, for example). Some vacationers to Missouri might have chosen shopping, restaurants or other forms of entertainment if casino gambling was not available. The Leven-Phares assumption that no out-of-state visitors to casinos in 1996 would have come to Missouri if they could not gamble neglects other reasons people come to the state. The result is an inflated impact estimate.

6. The Leven-Phares Study's Conclusion That Casinos Contributed a Net Benefit of $543 Million to the Missouri Economy in 1996 is Not Supported by Their Research.

          At a Minimum, the Net 1996 Benefit to Missouri Should be Reduced from the $543 Million Estimated by Leven-Phares to $361 Million, and the Introduction of Casinos Could Have Actually Resulted in a Net Loss to the State of $185 Million When Higher Plausible Estimates of Problem Gamblers are Considered. These Estimates Do Not Take Into Account Additional Losses Made Possible by Overestimates in Several Key Areas Noted Above.

          Leven-Phares estimated the net 1996 benefit to the state from casino gambling as $543 million. However, if the costs of problem gambling -- which they have omitted -- are taken into account, this alone would lower the net benefit level to a maximum of $361 million, if low-end estimates of problem gambling's cost impact are used, or to a net loss of $185 million, if higher plausible estimates of problem gambling's costs are considered. Furthermore, this does not include correcting for Leven-Phares' overestimates of in-state casino expenditures or recaptured spending by casino patrons. Correcting for such overestimates would generate an even smaller overall benefit estimate, at best, and an even larger net loss at worst.



          At the request of the Missouri Council on Economic Integrity, the United States Gambling Research Institute (USGRI) has undertaken this review of the study, The Economic Impact of Gaming in Missouri, by Charles Leven and Don Phares, prepared for Civic Progress of St. Louis, MO, and released in April, 1998 (the Leven-Phares study). Taking into account that the stated objective of that study's authors was to examine only economic impacts, USGRI has also restricted its review to economic impacts. We note, however, that there are also many social and political impacts that clearly need to be examined in any overall assessment of gambling in Missouri.

          Our basic objective in undertaking this review was to determine to what extent the Leven-Phares study provides a comprehensive understanding of the economic impacts of the Missouri gambling casinos on the Missouri economy.

          Unfortunately our ability to carry out a comprehensive review of the Leven-Phares study was hampered by the fact that the authors of the study would not give us access to the underlying data used in their report. Donald Phares, one of the co-authors of the study, declined a request by our economic researcher, Professor Edward Feser of the University of North Carolina, to provide this information, citing his promise to the casino companies that the economic data they provided to him would not be made public. In an e-mail to Feser, Phares wrote, "At the outset of this study it was absolutely necessary to assure confidentiality to the participants, which we did and must continue to stick by. I fully realize this makes your task more difficult but I am certain you appreciate that a guarantee such as we made must be honored."

          It is common to promise confidentiality to study participants and such promises must certainly be honored. But the decision by the Leven-Phares team is troubling since it is also common practice for researchers to allow freedom of access to their data so that other researchers can replicate and confirm results (while maintaining the confidentiality of the study participants). The lack of access to the study data left us unable to confirm the Leven-Phares analysis of the economic impacts of gambling on the Missouri economy. Indeed, with undisclosed data as the basis of the Leven-Phares report, there is no truly independent way for others to examine it. This is particularly disturbing in a report which the authors describe as providing an accurate empirical basis for debate on the casino issue.

          Although we outline a series of methodological problems with the Leven-Phares analysis, nevertheless their study does contain a good deal of useful information to help understand the economic impacts of casinos in Missouri. We hope that this review, read in conjunction with the Leven-Phares study, will enable Missouri legislators, public officials and citizens to have a fuller and more accurate understanding of these impacts.

Return to top of page

The Leven-Phares Study: Its Methodology and Findings

          In a report released in April 1998, Charles Leven and Donald Phares argued that in 1996 the seven casinos then operating in Missouri generated $543 million in new spending, $324 million in personal income, and over 12,100 jobs for the state's economy. Projections for 1997 (accounting both for new casinos opening in that year and anticipated growth in admissions and revenues) found the gambling industry creating $759 million in output, $508 million in personal income, and over 17,900 jobs. The study also found that in 1997 casinos provided $175 million in state revenues (78 percent of which were devoted to education spending), $63 million in revenues to home dock cities, and $51 million in annualized construction dollars. While gambling taxes and fees generate revenue that is a very small fraction of total state education spending in Missouri, the study showed that revenues provided by casinos constitute a significant share of total receipts for some home dock communities (from 75 percent in Riverside to 52 and 42 percent in Caruthersville and North Kansas City, respectively).

          The Leven-Phares results are based on the calculation of direct, indirect, and induced spending from casino operations, net of consumer spending that would probably have occurred in the absence of the casinos (what is termed the 'displacement effect'). The study therefore accounts for the multiple subsequent rounds of spending that occur when Missouri casinos pay out wages to casino employees, make purchases from vendors located in the state, and pay taxes and fees to state and local governments. It was carried out with an approximated Missouri input-output model (called IMPLAN) developed by Minnesota Implan Group, Inc. (MIG), secondary data from MIG, and primary data from surveys of casino patrons and operators. The surveys provided the critical information necessary to estimate in-state spending by Missouri casinos as well as the expenditures that Missouri patrons would have made locally (e.g., for basic needs, other forms of recreation, etc.) if they had not spent the funds in the casinos. The survey of casino patrons also provides a profile of the 'average' gambler in Missouri.

          To examine the accuracy of these conclusions, we examined the appropriateness of the methodology used by Leven-Phares. In our work we chose to focus on the 1996, rather than the 1997, impact analysis, since the 1996 analysis was based on actual data rather than projections. We sought to determine: Did the casino industry in 1996 actually produce $543 million in net direct, indirect and induced benefits to the public and private sectors of the Missouri economy?

Return to top of page

Strengths and Shortcomings of the Leven-Phares Methodology

          As a result of our examination, we determined that the Leven-Phares study exhibited certain methodological strengths, but was also marred by some methodological approaches that were of questionable validity. Among the strengths of the study was its basic objective, with which USGRI concurs: in the study's own words, "to take an objective look at the economic impact of the gaming industry on the Missouri economy," and to determine "the bottom line economic impact." Leven-Phares used an acceptable model for economic impact analysis, the IMPLAN input-output model, which accounts for the multiple rounds of spending that occur when casinos purchase goods from in-state vendors, pay their employees -- who then make purchases from vendors located in the state, and pay taxes and fees to state and local governments.

          The Leven-Phares study is among few that have been conducted to date that attempt to explicitly document displacement spending associated with casino gambling. Most studies adopt an 'either-or' approach, where all spending made by local casino patrons is considered either substitution of other spending (thus yielding no effect) or completely new spending (thus generating a very large impact). The Leven-Phares analysis attempts to find a middle ground, where some local spending is indeed 'new' to the Missouri economy since it represents either a depletion of savings or a recapture of spending that would have otherwise been made out of state (e.g., for trips to out-of-state casinos, vacations, catalog mail orders, etc.) while other local casino spending is 'old' in that it would have been made in Missouri stores, restaurants, and other businesses if casino gambling was unavailable.

          However, the Leven-Phares study suffers from two important methodological flaws that, in toto, result in a significant overstatement of the benefits of casino gambling to the Missouri economy. These are 1) a bias towards the overestimation of in-state benefits resulting from the casino industry, and 2) a complete failure to consider any of the substantial costs resulting from the increased problem gambling associated with the legalization and spread of casino gambling.

          The co-authors of this USGRI review each focused their attention on one of these two areas, consistent with their own areas of research expertise. The methodological issues associated with the input-output analysis undertaken by Leven-Phares are discussed in the following section of this report by Professor Edward Feser. After that, Professor Robert Goodman addresses the issues raised by the consideration of the costs associated with problem gambling.

Return to top of page

Assessment of the Methodology of The Economic Impact Analysis

Prepared by Edward Feser

          This section provides a description and evaluation of the impact analysis methodology behind the Leven-Phares study. It focuses on assessing the quality of the study and the likelihood that the Leven-Phares analysis provides valid estimates of the positive economic impact of gambling on the state. Whether or not the Leven-Phares's study constitutes an adequate picture of both the benefits and costs of casino gambling in Missouri is addressed elsewhere in this report. What is at issue here is the following: accepting input-output based economic impact analysis as one means of documenting the contribution of the gambling sector to Missouri's economy, how successful is the Leven-Phares' analysis in providing that documentation?

Our Approach
          A significant number of assumptions underlie the typical input-output based economic impact analysis. For example, the derivation of impact multipliers for most U.S. states requires the 'regionalization' of national-level input-output tables (meaning that state tables are approximated with baseline data for the U.S. as a whole). Primary data on the purchases and sales of commodities between sectors and industries is, with very few exceptions, available only at the national level. To develop local multipliers, analysts must select among a limited set of techniques for estimating local versus non-local purchases and sales between industries and among final demand sectors. Such techniques are naturally imperfect and therefore the use of them should not necessarily be beyond criticism. Indeed, debate over the best techniques for regionalizing national input-output tables is ongoing in the academic literature.

          However, the use of a regionalized input-output table (as is done in the Leven-Phares study) is necessary given the present state of economic impact methodology. Without regionalized tables, it is impossible to conduct a state or local economic impact analysis of this kind. And on the whole, researchers and academics have concluded that such tables can generate reasonable and defensible estimates. It is generally accepted in the research community that such estimates can establish an important basis for more informed policy discussions.

          Rather than evaluate methods and assumptions that are common to all input-output based impact analyses, this assessment concentrates on those methodological procedures adopted by Leven-Phares that are specific to the impact analysis of gambling in Missouri. This includes, among other things, the validity of inferences drawn from primary and secondary data and the approach to measuring the in-state spending displacement effect.

Leven-Phares Provide Only Limited Access to Their Data
          Ideally, to properly assess the Leven-Phares study we should carefully measure the influence of those procedures and assumptions that we question on the final results. In principle, we could accomplish this task by generating additional impact estimates under alternative (and presumably more plausible) assumptions. Assumptions that when altered slightly cause the greatest shifts in the impact estimates should be subjected to the highest degree of scrutiny to ensure they are both reasonable and valid. At a minimum, they should be highlighted so that the public is aware that any economic impact estimates are based on a host of key assumptions, most of which are hopefully substantiated by data but others which are based primarily on theory or even informed guesses. It is important to realize that it is possible for reasonable economists to disagree on what these assumptions should be.

          To conduct such an assessment, however, we must have access to both the models and data used in the original study. While IMPLAN software and baseline data are commercially available, we do not have access to most of the raw data underlying the Leven-Phares analysis. As noted above, the Leven-Phares research team undertook a considerable data collection effort that involved soliciting detailed information from seven Missouri gambling houses as well as administering a 74 question survey to 823 casino patrons. Unfortunately, our request to view these raw data was denied.
2 Missouri casinos were assured in the course of the study that no data they provided would be disclosed to the public. Therefore, the Leven-Phares team believes that it must not release the data, even for an independent assessment conducted in a manner that the information would remain confidential.

          The implication of this is that while we are able to identify several questionable assumptions and procedures in the original study, we cannot verify the precise influence of those assumptions on the final impact estimates. In most cases we are able to determine whether the assumptions might have led to the overestimation of the positive impact of gambling on Missouri's economy. In other words, while we can speculate on the likely direction of the influence of a particular assumption on the Leven-Phares results, we cannot produce precise revised estimates for all factors. The latter would be impossible without a full data collection effort on the scale of the original study. Sometimes we can reasonably speculate on the magnitude of the influence but the scope of our assessment is clearly limited.

          We readily admit that some of the issues we raise may simply be a function of inadequate documentation of research procedures in the published report and its appendices. We expect that some of our questions can be answered by the Leven-Phares team with a fuller explication of their research procedures. In that case, public dialogue in Missouri over the results of the Leven-Phares study will be enhanced by making clearer the basis of their findings.

          We hope that future researchers studying the economic and social impacts of the gambling industry will commit to making their data and methods fully available to outside reviewers interested in verifying their results. As Leven and Phares note, in almost any other sector of the economy, the results of an impact analysis would inspire little controversy. In the case of casino gambling, the very controversy it inspires emphasizes the need for full and frank discussion of the positive and negative impacts of gambling on local economies. This necessitates an exceptional level of commitment by researchers to a process of independent review and verification.

Issues Raised by the Leven-Phares Methodology
          It should be stated at the outset that our review of the methodology in the Leven-Phares study indicates that the analysis, while very limited in its primary focus on the benefits of casino gambling in Missouri, is fundamentally sound within the context of best-practice economic impact analysis. The research team use a well-accepted input-output model to measure the total impact (direct, indirect, and induced) of dollars spent for casino operations (including labor and non-labor inputs), taxes and fees paid to the state, taxes and fees paid to home dock cities, and expenditures for construction. Spending by casino patrons that would have been spent in Missouri in the absence of the casinos is estimated and deducted from total impact while dollars contributed from savings or from 'recaptured' spending are accounted for.

          While the general approach is sound however, there are some problems associated with the implementation of the methodology

Estimation of Payments to Missouri Laborers.
          There are two major elements to the estimation of total impacts from operational expenditures by the seven casinos in Missouri: labor and non-labor inputs. The casinos reported $131.1 million in total labor expenses in 1996. An important question is what share of those expenditures on salaries and wages accrues to workers that are Missouri residents. Non-Missouri residents would be expected to make many of their purchases in their home communities, denying Missouri of some of the additional rounds of induced spending that derive from the initial wage payment. The information on local versus non-local wage payments was actually requested from the casinos themselves. Based on that self-reporting, Leven-Phares estimate that $124 million in wages and salaries in 1996 was paid to Missouri workers, or 95 percent of the total. After netting out social security taxes and fringe benefit costs that cannot be regarded as local (within the state of Missouri), Leven-Phares calculate the total impact of $113.5 million in wage and salary payments.

          If the percentage of Missouri casino workers is actually lower than 95 percent, then total labor impacts (estimated at $200.6 million in 1996) are overestimated. Is there any reason to believe that the share is lower? In fact, all eleven Missouri casinos (seven operational since 1996 and four opening subsequently) are located in Missouri border counties. Six casinos are located in Kansas City area, a Metropolitan Statistical Area (MSA) that includes seven Missouri and four Kansas counties. The counties of an MSA typically comprise, according to rough Bureau of Census definitions, a labor market area or region within which most residents also hold their jobs. It is certainly possible that casino workers could live in Kansas and work in Missouri. Likewise, the St. Louis MSA includes four Missouri and five Illinois counties. The four casinos within the St. Louis area would pull from this larger two-state labor market.

          In Section X of their study (p. 68-9), the Leven-Phares team acknowledges that some casino workers may commute to Missouri from neighboring states. However, they argue that accounting for this problem is necessary primarily for local-level studies (e.g., a study of the impact of the gambling industry restricted to Kansas City). However, this logic is tenuous at best since if there is a significant degree of commuting at the local level, statewide impacts would likewise be affected. In addition, use of self-reported data by the casinos without some independent verification is questionable since casinos have a incentive to over-estimate the share of their workers residing in Missouri.

Estimation of Non-Labor Expenditures.
          The other element of total casino operation expenditures is payments to vendors for supplies and equipment. Leven and Phares report those payments for 1996 at $237 million, 92 percent of which are made within Missouri. The source of local to non-local spending ratio is again self-reporting by the casinos themselves although unlike the case with labor costs, the data collection instrument sent to the casinos (provided in the report appendix) does not ask respondents to distinguish between local and non-local spending. Again, even if casino operating expenses involve mostly purchases that are generally made locally (e.g., professional services, insurance, management assistance, legal services, etc.), the fact that most casinos are situated in metropolitan areas that span state borders would suggest that the share of Missouri spending may be lower than 92 percent. But it is also important to note that most casinos in Missouri are owned by national gambling companies that are headquartered outside of the state. Non-locally owned companies typically source a higher share of their inputs from non-local vendors. If the share of local purchases is lower than 92 percent for these reasons, the total impact of non-labor expenditures is over-estimated. Unfortunately, the report provides insufficient documentation of how the 92 percent share was determined other than to cite surveys to casinos.

Estimation of Displacement Effect: Out-of-State Visitors.
          One of the most important aspects of the study is the estimation of Missouri consumption spending that is displaced as Missouri residents shift some of their expenditures to the gambling sector. That 'displacement effect' is an offset to the total impact of casino spending for labor and non-labor costs, taxes and fees, and construction. Based on data collected from the casinos, Leven-Phares determine that out of the $589 million in spending at the casinos in 1996, 30 percent came from out-of-state visitors, 17 percent came from reductions in out-of-state spending by Missourians, 8.3 percent came from depletion of savings, and 45 percent came from displacement of Missouri expenditures in other sectors and industries.

          One problem with the displacement effect estimates by Leven-Phares relates to the out-of-state visitor spending. Leven-Phares accept a self-reported estimate from the casinos regarding the share of funds derived from out-of-state visitors ($177 million). They then assume that the full $177 million represents new expenditures in Missouri (funds that would not be spent in the state in the absence of gambling casinos), noting that "only incidently would such spending have gone to other businesses in Missouri, so that it can safely be assumed that all of the spending by non-residents is net new spending in the state. Even anti-gaming interests often concede this point" (p. 31-2).

          The problem with this assumption is that it does, in fact, matter whether such spending would occur incidently. Imagine a case in which an out-of-state tourist travels to Missouri for reasons that are totally unrelated to casino gambling yet chooses to go to a casino while in the state. In that instance, in the absence of the casino, that tourist may have visited a sporting event, restaurant, theater, movie, or other type of entertainment in Missouri. In other words, it is unlikely that all out-of-state casino patrons travel to Missouri primarily to visit casinos. Moreover, it is standard practice in tourist site impact analyses to conduct on-site visitor surveys to determine the principal reasons for visitor trips to given tourist attractions. By doing this, researchers are able to determine if sites are national, regional, or local draws.

          Las Vegas, for example, attracts gambling patrons from throughout the U.S. and even world. Las Vegas casinos are, in effect, a national tourist draw. Smaller gambling areas such as those on India reservations and in places like Illinois, Mississippi, Missouri tend to attract visitors locally and regionally. Other things equal, a higher share of out-of-state visitors to regional -- as opposed to national -- attractions are probably in the gambling state for other reasons (other tourist attractions, to visit family or friends, etc.). For visitors in the state mainly for reasons other than gambling, the visit to the casino is incidental to the other business of the individual or family and thus may be displacing spending that would have been made in other sectors. Neglecting the displacement spending from out-of-state casino patrons of this type overstates the overall casino spending impact.

Estimation of Displacement Effect: In-State Spending Offset
          As noted above, 45 percent of Missouri casinos' 1996 expenditures are considered displacement of other types of spending that would have been made within Missouri. Leven-Phares essentially use survey data to estimate what Missouri patrons would have spent on out-of-state vacations, out-of-state horse and dog tracks, out-of-state gambling, and a variety of in-state expenditures (recreation, basic needs, etc.). The avoided out-of-state expenditures constitute dollars 'recaptured' for Missouri by the casinos. The miscellaneous in-state dollars represent a offset to other casino spending. In principle, these procedures are reasonable. But there are several problems with their implementation that could have led to an underestimation of the displacement effect and therefore a overestimation of the overall impact of gambling houses on the Missouri economy.

          First, the sample frame for the 74 question survey from which displacement effects are estimated is probably not representative of the gambling population in Missouri. In fact, frequent gamblers are likely to be over-represented in the results. The Leven-Phares team only surveys casino patrons that visited Missouri casinos a minimum of three times in 1996. Indeed, based on the survey results reported in the appendix, the 823 respondents to the survey visited Missouri gambling houses an average of 17 times in 1996.
4 Frequent gamblers are much more likely to report traveling to out-of-state casinos as well as horse and dog tracks when casino gambling was unavailable in Missouri. If the findings for the frequent gambler are extrapolated to all Missouri gamblers the volume of total dollars recaptured by the casinos will be over-estimated.

          Second, it is unclear in the Leven-Phares report and appendices just how recapture and displacement dollars were estimated. Leven and Phares state the following: 'we asked Missouri casino patrons how much they spend on out-of-state casinos, out-of-state horse and dog tracks, and out-of-state vacations before there were casinos; what they recalled spending in 1996; and what they think they would have spent had there been no casinos (p. 36).' They go on to note that these questions produced 'highly consistent' estimates of the recapture dollars.

          A review of the survey (provided in the report's Appendix A) does not support these claims. In fact, although the Leven-Phares team did ask several closely related questions aimed at deriving information about recapture and displacement spending, only one set of questions actually requested that respondents provide dollar amounts. For example, survey question 24 asks 'Before gaming was available in Missouri, how many more times per year did you visit a casino in another state than you do now?' Duplicate questions 25-31 request total visits to spectator sporting events, movies, the theater, bingo, restaurants, and participant sporting events. Question 34, re-states 24 in a slightly different way: 'If gaming had not been available in Missouri in 1996, do you think you would have gone out of state to visit a casino or gaming activity more often than you did?' Question 35 follows up with: 'How many times more often would you have gone out of state to visit a casino or gaming activity.' Finally, question 36 requests dollar figures on such activity: 'About how much more money would you have spent on these extra visits in 1996.' Questions 37-61 essentially repeat the three-question sequence for each category of spending in questions 25-31, using slightly different wording. No other question about recapture or displacement expenditures is asked of the respondent.

          If the results from parallel questions in the sets 24-31 and 34-61 are compared, one finds that there is actually little consistency in the findings. For example, to the question 'Before gaming was available in Missouri, how many more times per year did you attend a spectator sporting event than you do now (question 25),' 190 respondents (23 percent) answered that they would have done so one or more times. To the slightly reworded question 37 (which follows the question 'If gaming had not been available in Missouri in 1996, do you think you would have attended a major sporting event more often than you did?') which asks 'How many more times would you have attended a major sporting event?,' all respondents answered zero. This significant discrepancy in findings between the two parallel sets of questions holds for all other spending categories.

          An additional important point to note is that the survey does not actually specify spending for out-of-state vacations. Questions 58 and 59 address vacation activity. Question 58 reads, 'If gaming had not been available in Missouri in 1996, do you think you would have taken more vacation than you did?,' while 59 reads, 'About how much more money would you have spent on your vacation in 1996?' Leven and Phares report that $16 million was recaptured for Missouri by casinos that can be attributed to Missouri residents remaining home to gamble rather than spending their money on out-of-state vacations. However, the survey does not permit the distinction between out-of-state and in-state vacation expenditures. Therefore, the $16 million presumably includes some spending that is recaptured to the state but is rather displaced from other sectors by the casino industry. Leven-Phares may have adjusted the survey estimates in some way to account for in- versus out-of-state vacation spending, but if they did, they did not document it in the report.

Estimation of Construction Spending Impacts.
          The Leven-Phares team relies on data from seven Missouri casinos for estimates of total construction spending and the share of it that is spent within the state. As we have already noted, relying on industry-reported data can be suspect, especially when it does not appear that the numbers were verified with independent information. Nonetheless, the data Leven-Phares report indicate that of the $534 million spent on total construction-related expenses, about 64 percent ($340 million) occurred in Missouri. While this estimate is more reasonable than the local shares of labor and non-labor spending, the fact that six of the seven casinos that supplied construction data for the Leven-Phares study are located in Kansas City or St. Louis, both cities situated on the Missouri border, additional documentation of how the figure was arrived at is necessary. As in the case of operating expenses, there is little discussion of the estimation of local spending by the casinos other than references to casino-supplied information.

          The implications of the Leven-Phares annualization of construction impacts is also worth noting. Since construction activity offers only short-term (12-18 months) economic benefits, standard practice in economic impact analysis is to report construction impacts for only the first year of a project horizon. Leven and Phares annualize the impacts so that they can come up with an overall benefit figure that is not distorted by huge, one-time construction impacts. While not intended to be misleading, this is an unorthodox procedure that may confuse policy makers.

          If one is interested in the impact of ongoing casino operations in the state, one should include only activities related to daily casino operations. Of the $543 million in casino impacts for 1996 reported by Leven-Phares, $59.3 million are associated with annualized construction impacts. Policy makers must understand that the actual annual impacts once casinos are up and running are closer to $484 million.

          In general, spending should be modeled separately for a time horizon equal to the length of the construction project itself. Casinos will not generate nearly $60 million in construction impacts over the next twenty years. Rather, they will generate a large construction impact in the first twelve to eighteen months, leading to a brief increase in output, income, and employment in the state. Once the construction is completed, those impacts disappear. Annualizing the construction numbers and including them in the operating impact figures only obscures the fundamental differences in the types of spending. It is too easy for this to lead to confused policy discussion.

Return to top of page

Assessment of the Issues Raised by the Economic Costs of Problem Gambling

Prepared by Robert Goodman

          The authors of the Leven-Phares study made the decision explicitly to exclude "some economic or economic-related issues which are beyond the scope of this analysis." One such excluded economic issue was the cost of problem gambling. In their report, Leven-Phares justified their decision to exclude consideration of problem gambling costs on the grounds that such costs were 1) difficult to determine and 2) likely to be small. Unfortunately, neither of these premises upon which Leven-Phares based their exclusion of problem gambling costs is correct.

Can Problem Gambling Costs Be Omitted?
          The lack of cost estimates of the impact of problem gambling resulting from casinos in Missouri in the Leven-Phares study was also raised during a presentation of this study to the congressionally-mandated National Gambling Impact Study Commission. In hearings before the Commission on May 20, 1998, Don Phares was asked the following by Commissioner Leo McCarthy, a member of the Commission's research committee:

[M]y question to you is if you had broad latitude in designing the study and you knew the social cost data were lacking, and indeed they are because we've been searching for it continuously, why weren't questions on social costs included in your study to give a balanced picture of the cost to government, the cost to the private sector employers for pathological gamblers?.. . it's a reasonable assumption that lots of different kinds of costs come out of that [problem gamblers]. My question is why weren't any social costs, factors, questions included in your study ? [our emphasis].

          Phares responded to Commissioner McCarthy's question by saying that there were three reasons for not including such cost data in his report: 1) that it was too difficult to isolate the cost to society from the problem gambler from the cost borne by the problem gambler him or herself , 2) the lack of time to do the work; and 3) the lack of budget to do the work. According to Phares:

   What is a social cost and what is an individual cost requires a lot more empirical examination. What is a problem gamer versus someone who spends a large percentage of their money on entertainment as gaming is a question which has not been addressed. Is it $5,000 for a problem gamer or $105,000? What is the value of a human life? Why does someone commit suicide?

   One of the reasons why is that this issue is simply so wide open that it would have led into the second two issues. One was a question of time. We had a time frame in which to do this. And third was a budget. To do what you are suggesting would have required a budget several times what we had to do the study. . . . It should be done in Missouri. It has not been done in Missouri, for basically the reasons I indicated [our emphasis].

          These were not the reasons given for the exclusion in the Leven-Phares study itself. Before the Commission, Phares chose not to use the reason he gave in the report: the assertion that "it seems unlikely that the economic burden [of problem gambling] on the state is a large problem" [our emphasis].

          In their report, Leven-Phares recognized that the costs of problem gambling did have a long-term economic effect on both the public, and private sectors. "The most difficult part of estimating long-run impacts on government and non-profit finances," they wrote, "center on estimating additional demands for and costs of welfare services generated by casino gambling. At the heart of this problem is the lack of reliable data on the extent of problem gambling" [our emphasis] (pp. 71, 72). Rather than attempting to determine the relevant data, the authors simply made the assumption that the introduction of legalized casino gambling was likely to have little effect on the prevalence of problem gambling, and that such problems would exist at the same level with or without the introduction of casinos. As they stated: "[M]uch of the problem might simply be a substitute for other forms of 'problem gambling.' Or maybe it is a substitute for other forms of self destructive behavior. . . . the fundamental behavioral research needed to assess the net socio-psychological impacts on society just has not been done. . . . we note that available studies suggest that only a small fraction of gaming patrons seem so affected. . . . it seems unlikely that the economic burden on the state is a large problem" (p. 72).

          But these conclusions about the incidence of problem gambling appear to be the unsubstantiated assumptions of the researchers, and are not warranted by existing empirical evidence. The authors give no specific explanation of why they reached these conclusions. Although they do say that their survey sample of casino patrons "suggests that much of the problem might simply be a substitute for other forms of 'problem gambling'," they provide no analysis to support such a conclusion. In fact, there are no questions at all in their survey that ask about problem gambling.

Research Indicates Problem Gambling Costs Are Significant
          The Leven-Phares report ignores the extensive research that has been done in the field of problem gambling. This previous research indicates a strong connection between increases in the availability of legalized casino gambling and increases in problem gambling. Leven-Phares appear to reject or be unaware of any of the important empirical studies of the economic costs associated with increased problem gambling, which include the 1985 work of Robert M. Politzer, James S. Morrow, and Sandra B. Leavey, "Report on the Cost-Benefit / Effectiveness of Treatment at the Johns Hopkins Center for Pathological Behavior," in the Journal of Gambling Behavior, Vol.1, No.2, Fall/Winter 1985; Rachel Volberg's 1993 economic impact analysis, "Assessing the Social and Treatment Costs of Gambling," presented at the Seventh National Conference on Gambling Behavior, July 22-24, 1993, New London, CT, National Council on Compulsive Gambling; William N.Thompson, Ricardo Gazel, and Dan Rickman's study, The Social Costs of Gambling in Wisconsin, Wisconsin Policy Institute, Thiensville,WI, April 1996; and the Harvard Medical School 1997 Problem Gambling Study by Howard J. Shaffer et al., "Estimating the Prevalence of Disordered Gambling Behavior in the United States: A Meta-Analysis," Harvard Medical School, Division on Addictions, Boston, MA, December 15, 1977.

          Even if it were true, as the Leven-Phares study states, that "only a small fraction of gaming patrons" are problem gamblers, this can still represent a enormous economic burden on the rest of the people in a state. For example, the U.S. Department of Health and Human Services determined in 1991 that only 1.88 percent of the U.S. population had a dependency on illicit drugs. But the agency also estimated that the year before, this small fraction of people with drug problems cost the country $66.9 billion in terms of health costs, lower productivity, and higher crime.
5 According to the Department, each of the 4.7 million drug dependent persons in the country contributed an average of approximately $14,200 to that total cost. Similarly, the social costs of each individual problem gambler are high, and even small increases in the proportion of problem gamblers in a state can be extremely costly to that state's citizens.

          Thus, one critical conclusion of the Leven-Phares study, the assertion that "it seems unlikely that the economic burden [of problem gambling] on the state is a large problem," is neither substantiated by their research nor by that of other researchers in the field. In fact, quite the opposite has been shown to be true: the available research indicates that problem gambling is likely to have a major economic impact on the state of Missouri.

          First, there is a general consensus among researchers in the field that, contrary to Leven-Phares' assumption, increased availability of gambling opportunities does lead to increases in problem gambling behavior. Even research funded by the gambling industry supports this conclusion. The 1997 survey of the prevalence of problem gambling in the United States and Canada by a group of researchers at the Harvard Medical School's Division of Addictions, prepared for the National Center for Responsible Gaming, a casino industry-funded research institute in Kansas City, Missouri, found over one hundred relevant studies of problem gambling prevalence done between 1977 and 1997.
6 More than half of these studies were produced since 1992. The Harvard study refutes the idea that there is a lack of empirical evidence about the prevalence of problem gambling. While some problem gambling experts believe the study underestimates the level of problem gambling in the country, the casino industry itself has accepted the study as an index of the extent to which problem gambling exists.7

          This Harvard Medical School Study found a clear connection between the availability of legalized gambling opportunities and the rise in problem gambling. According to the study: "Newly exposed to the gambling experience, adults in the general population are having difficulty adjusting and, unlike the other population segments who are already evidencing gambling problems, are beginning to report increasingly higher rates of gambling disorder."

          The Harvard Study demonstrated that there has been a major increase in the prevalence rate of what they describe as "level 2" and "level 3" problem gambling behavior since the widespread introduction of legalized casino gambling in the early 1990s. Level 2 gambling, according to the report, "represents a wide range of adverse reactions." Level 3 gambling is described as having "the most serious" negative consequences.

          According to the study, the mean adult lifetime prevalence rate of combined level 2 and level 3 problem gambling had increased from 4.38 percent of the adult population during the 1977-1993 period to 6.72 percent of the adult population during the 1994-1997 period, a more than 50 percent (53%) increase. Level 3 gambling alone increased by more than 50 percent (54%); from 0.84 percent in the 1977-1993 period to 1.29 percent in the 1994-1997 period.
8 Since most of the new casino gambling was introduced in the United States and Canada in the early 1990s, and since gambling problems can take several years to manifest themselves, this increase in the prevalence of problem gambling is consistent with the generally accepted view that increased access to gambling leads to increased problem gambling.

          Secondly, the available research also indicates that increases in problem gambling result in significant increases in economic costs to society. For example, research from Louisiana shows that pathological gamblers spend an average of 45 percent of their monthly income on gambling expenditures.
9 Other researchers and problem gambling counselors have described the economic cost to society that results from problem gambling behavior.10 This includes the inability of problem gamblers to pay off debts they have contracted as a result of their gambling, as well as their failure to pay their taxes, utility bills and other monies they owe. Those who do pay their debts often get the money through criminal activities which include writing bad checks, engaging in fraud, embezzling money, dealing drugs, and simply stealing. The unpaid debts and bankruptcies of problem gamblers are not economic losses simply to the gamblers themselves, but also to those to whom the money is owed. Similarly, the criminal activities of problem gamblers translate into economic losses to those victimized and into increased costs to the taxpayers for police, courts, probation officers, and other aspects of the criminal justice system. The public also picks up the cost of keeping convicted offenders in jail. These incarceration costs can range from $20,000 to $50,000 a year, depending on the age and health of the prisoner.

          Problem gamblers are also costly to their employers, since they are frequently inattentive and unproductive at work. According to Dr. B. Kenneth Nelson, assistant director of psychiatry at Valley Forge Medical Center and Hospital, "Even where outright crimes have not been committed and/or detected, frequent absenteeism, tardiness, and squandering of company time and resources can add up to sizable financial losses for business."

The Costs of Unpaid Debts to a State's Economy
          A University of South Dakota study reported in 1991 that state-wide, "Chapter Seven bankruptcy filings and small claims filings have experienced significant increases in the two fiscal years since gaming began in South Dakota."
12 In 1995, four years after the introduction of riverboat casinos in Iowa, for example, Tom Coates, executive director of the Consumers Credit Counseling Service of Des Moines, stated, "We're seeing 15 to 20 percent of our counseling today has gambling as an important factor in reaching financial problems. Ten years ago, it was 2 to 3 percent.13

          In May 1997, the SMR Research Corporation produced the first national study on the relationship between bankruptcies and gambling. The study compared 1996 aggregated bankruptcy rates in counties with no gambling facilities with rates in counties having one or more gambling facilities. According to the SMR study, "The bankruptcy rate was 18 percent higher in counties with one gambling facility and it was 35 percent higher in counties with five or more gambling establishments." The authors noted that these figures actually understated the problem since many counties with gambling facilities had small populations and the players came from elsewhere. "[W]hen we look only at counties with more sizeable populations and gambling facilities, we see even greater evidence of the problem."

          Clark County, Nevada, where Las Vegas is located, had the state's highest bankruptcy rate; the two California counties just across the border from Las Vegas had the highest bankruptcy rates in California. Atlantic City, NJ, home to the Atlantic City casinos, had a bankruptcy rate more than a 70 percent higher than the state average, and most of the counties closest to Atlantic City had higher bankruptcy rates than those further away. Some of the highest bankruptcy rates in America are in the counties in the tri-state area of northern Mississippi, western Tennessee and eastern Arkansas that are near the major casino center of Tunica, MS. This area included Shelby County, TN, the location of Memphis, which in 1996 had the highest bankruptcy rate in America.

The Economic Costs of Increased Crime
          Cities that have introduced casinos have witnessed dramatic increases in crime and social problems. In Gulfport and Biloxi, Mississippi, every category of crime increased the year following the opening of casinos. Murders, rapes, robberies and car thefts more than doubled.
16 Domestic violence rose almost 70 percent on the Mississippi Gulf Coast after casinos opened there, and child abuse and neglect case filings increased by 15 percent in the first year of casino operation in Deadwood, South Dakota.17 A 1989 crime impact study of the Atlantic City casinos found that crime spilled over into the communities surrounding Atlantic City.18

          Studies indicate that the people who engage in crime in order to support compulsive gambling behavior generally have no prior record of criminal behavior. May ordinary people, without either criminal backgrounds or criminal inclinations, may engage in criminal activities when casino gambling is available.
19 In 1994, Jeffry Bloomberg, the State's Attorney for the town of Deadwood, South Dakota, told a congressional committee, "We have seen individuals who, prior to their exposure to gambling had no criminal history, who were not junkies or alcoholics, many of whom had good jobs, who became hooked on slot-machines and after losing all their assets and running all credit resources to their maximum began committing some type of crime to support their addiction." Bloomberg's statistics supported his description of increased numbers of child abuse and neglect cases seen by his office.20

Estimating the Costs of Problem Gambling in Missouri
          We can estimate the economic impacts of increased problem gambling resulting from casinos in the state of Missouri by first estimating the likely increase in the number of problem gamblers in Missouri since casinos were introduced into the state, and then applying a estimate of the average cost to the public and private sectors of the state for each of these problem gamblers:

Increased Problem Gamblers in MO Since Riverboats X Cost /
per Problem Gambler / per Year =
Total Cost of Problem Gamblers to MO / per Year

          Since Leven-Phares did not conduct research on the incidence or costs of problem gambling specifically in Missouri, we can extrapolate from the existing empirical studies of problem gambling impacts in other states to provide a measure what these impacts are likely to be in Missouri. Because analyses elsewhere cannot be taken as precise guides to the economic costs of problem gambling in Missouri, we use them to determine a range of probable impacts, from a low to a high estimated benchmark.

          The studies upon which we are relying were prepared by recognized experts in the field:
Howard Shaffer, the Director of the Center for Addiction Studies at Harvard University; Rachel Volberg. president of Gemini Associates in Northampton, MA and arguably the leading researcher on problem gambling prevalence studies in the country; William Thompson, professor of Public Administration at the University of Nevada-Las Vegas (UNLV); Ricardo Gazel, a former associate director of the Center for Business and Economic Research at UNLV and currently a researcher at the Federal Reserve Bank in Kansas City; and Dan Rickman, a professor of economics and business at Oklahoma State University.

          We used two sources to arrive at figures for the probable increase in the prevalence of problem gambling in the state of Missouri as a result of the legalization of the casinos. Since no prevalence study has been done in Missouri, we have relied for one estimate of this increase on the findings of what is probably one of the few benchmark studies of problem gambling related to the introduction of riverboat casinos in a mid-Western state -- the 1995 prevalence study commissioned by the Iowa Department of Human Services and produced by Rachel Volberg. This study, based on before and after surveys, determined that the rate of pathological or severe problem gambling increased after the introduction of riverboats in that state in the early 1990's from 0.10 percent of the adult population to 1.9 percent of the adult population -- an increase of 1.80 percent. (This study also determined that the combined rate of both problem and pathological gambling in Iowa more than tripled from 1.7 percent of the adult population before casinos to 5.4 percent of the adult population in 1995).

          A second source for our estimates is the Harvard Medical School Study, which surveys changes in problem gambling in the United States and Canada. We have used only the data in that study that apply to "Level 3" -- or the most severe level -- of problem gamblers, and we have also used only previous year prevalence rates, rather than life-time or total prevalence rates, since these yield the most conservative figures. Finally, we have used only adult problem gamblers for our estimates, despite the evidence of a growing number of youth problem gamblers. Another way of stating this is that we are estimating only the most immediate time frame for adults with the most extreme form of problem gambling.

          According to the Harvard Study, the average incidence of "previous year Level 3" problem gambling during the 1977-1993 period -- before the widespread introduction of legalized gambling in the U. S. -- was 0.84 percent of the adult population, while during the 1994-1997 period -- after the introduction of gambling -- it rose to 1.29 percent. This represents a net increase of 0.45 percent of the adult population.

          Thus, the increase in serious problem gamblers that can be reasonably attributed to the introduction of casinos in Missouri would range from a low of .45 percent to a high of 1.8 percent.

          To estimate the cost per problem gambler, we use a 1996 Wisconsin survey by Professors Thompson, Gazel and Rickman. It is important to note that their study produced the most conservative estimate of problem gambling costs to date, and therefore may err on the low side. They based their cost figures on a survey of problem gamblers, and produced separate figures for problem gamblers generally and for casino problem gamblers in particular. They estimated that the average serious problem gambler cost the public and private sectors $9,469 per year. The average serious casino problem gambler was estimated to cost the public and private sectors $10,113 per year. Using the Thompson, Gazel, Rickman figure means that each additional 100 serious casino problem gamblers in a state cost that state's economy more than one million dollars per year.

Estimated LOW and HIGH Costs of Serious Problem Gambling in Missouri in 1996

Increase Problem Gamblers in MO Since Riverboats X Cost per Problem Gambler per Year = Total Cost of Problem Gamblers to MO per Year

Estimated Increase in Percent of Serious Problem Gamblers in Missouri from 1991-1996:
          LOW: 0.45 percent
          HIGH: 1.80 percent

Estimated Yearly Cost of Each Serious Problem Gambler in Missouri: $10,113

1996 Missouri Adult Population: 3,997,000

Net Increase in Adult Serious Problem Gamblers in Missouri between 1991 and 1996:
          LOW ESTIMATE: 0.45% X 3,997,000 = 17,990 Serious Problem
          HIGH ESTIMATE: 1.80% X 3,997,000 = 71,946 Serious Problem

Total 1996 Cost of Problem Gamblers in Missouri:
          LOW ESTIMATE:  17,990 X $10,113/ year = $181,932,900, or
          $182 million

          HIGH ESTIMATE:  71,946 X $10,113/ year =$727,589,900, or
          $728 million

Return to top of page

          These estimates show that the costs of problem gambling in Missouri in 1996, using conservative estimates of the probable increase in only the most severe problem gambling, likely range from $182 million to possibly as high as $728 million. Thus, the Leven-Phares study's conclusion that casinos contributed a net benefit of $543 million to the Missouri economy in 1996 must be revised to reflect the costs of problem gambling. Once these costs are subtracted, the Leven-Phares research would suggest at best a net benefit to the state of $361 million and at worst a net loss to the state of $185 million. And once the probable overestimates of economic benefits pointed out by Professor Feser are calculated and taken into consideration, the net benefit would be smaller and the net loss even higher.

          While the Leven-Phares study's general economic impact methodology is fundamentally sound and even commendable in some respects, certain aspects of the implementation of the analysis (particularly in the areas of data collection and interpretation) are likely to have led to the overestimation of the benefits of casino gambling to the state's economy. Moreover, the complete failure to include any calculation for the costs of increased problem gambling, despite substantial research indicating the existence of such effects, significantly distorted the study's conclusions. Correcting for the study's bias towards overestimated benefits as well as the omission of problem gambling costs would lead to a correct figure for casino gambling's impact on the state's economy that would show, at best, a much smaller net benefit, and at worst a significant net loss.

Return to top of page

             1 E-Mail, Donald Phares to Edward Feser, September 24, 1998

          2 Frequency tabulations of the results of the patron survey are provided in one of the report's nine appendices. While these provide a considerable amount of information from the survey, the raw data are needed to fully assess the Leven-Phares methodology.

          3 Another potential problem with the estimation of non-labor casino expenditures is the inclusion by Leven-Phares of $37.3 million in depreciation as an operating expense. Accounting depreciation would not normally be considered a direct outlay in an impact analysis of this kind.

             4 Our best estimate of median visits is six to eight based on the frequency tabulation provided in the appendix.

             5 Beatrice A. Rose, Ed., Substance Abuse and Mental Health Statistics Sourcebook, Substance Abuse and Mental Health Services Administration, Public Health Service, U.S. Department of Health and Human Services, U.S. Government Printing Office, 1995. See pp.3 and

6 Howard J. Shaffer et al., Estimating the Prevalence of Disordered Gambling Behavior in the United States: A Meta-Analysis, Harvard Medical School, Division on Addictions, Boston, MA, December 15, 1997.

7 See Brett Pulley, "Compulsion to Gamble Seen Growing," New York Times, December

             8 Howard J. Shaffer et al., op. cit.

9 Westphal, J.R. and J. Rush, "Pathological Gambling in Louisiana: An Epidemiological Perspective," Journal of Louisiana State Medical Association 148 (1996), 353-358.

10 See, for example, descriptions in Arnold and Sheila Wexler, "Facts on Compulsive Gambling and Addiction," New Jersey Alcohol/Drug Clearing House, Center for Alcohol Studies, Rutgers University, Piscataway, NJ (1992) and Henry R. Lesieur, "Compulsive Gambling," Society (May/June 1992)

              11 Nelson, B. Kenneth, "Not Just A Game," Focus (Philadelphia Business Weekly), May 16, 1990.

12 Madden, Michael K., R.A. Volberg & R.M. Stuefen, Gaming in South Dakota: A Study of Gambling Participation and Problem Gambling and a Statistical Description and Analysis to its Socioeconomic Impacts, Business Research Bureau, University of South Dakota: Vermillion, SD (November 12, 1991).

          13 Santiago, Frank, "Betting on Bankruptcy?" Des Moines Register, December 3, 1995.

          14 SMR Research Corporation, The Personal Bankruptcy Crisis, 1997: Demographics, Causes, Implications, and Solutions, Hackettstown, NJ, 1997.

             15 Id.

             16 Report of Maryland State Attorney General J. Joseph Curran, Jr., The House Never Loses, and Maryland Cannot Win: Why Casino Gambling is a Bad Idea, October. 16, 1995, p. 21.

          17 Report of Maryland Attorney General, supra, pp. 23, 32.

          18 Friedman, Joseph, Simon Hakim and J. Weinblatt, "Casino Gambling as a 'Growth Pole' Strategy and Its Effect on Crime," Journal of Regional Science, Vol. 29, No. 4 (1989), pp. 615 - 623.

          19 Jacobs, Durand F., "Problem Gambling and White Collar Crime," paper presented at Seventh International Conference On Gambling and Risk Taking, Reno, Nevada (August 23-26, 1987), and Bloomberg, Jeffry L., Testimony at Hearing on the National Impact of Casino Gambling Proliferation, Committee on Small Business, U.S. House of Representatives, Washington, D.C. (September 21, 1994).

             20 Bloomberg, Jeffry L. Testimony before U.S. House of Representatives, supra.

          21 Gemini Research, Gambling and Problem Gambling in Iowa: Report to the Iowa Dept. of Human Services, Des Moines, Iowa (July 28, 1995).

Return to top of page