|
Article Excerpt
INTRODUCTION
Beginning with New Hampshire in 1964, 42 states and the District of Columbia have legalized state-sponsored lotteries. Lottery sales in the United States topped $48 billion in fiscal year 2006 (roughly $160 per capita), of which state governments retained nearly $17 billion (about one percent of total state government revenue) for spending on education, infrastructure, and other social programs. The preceding sales and tax revenues suggest an average tax rate of 35 percent, an average rate much higher than that of other state excise taxes (Clotfelter and Cook, 1987). The growth in lottery sales over the past several decades is not only a result of more state lotteries, but also of an ever-evolving product line that is designed to attract and retain customers through higher jackpots, some reaching several hundred million dollars.
The growth of the lottery industry has sparked much research. Numerous studies, including Filer, Moak, and Uze (1988), Davis, Filer, and Moak (1992), and Alm, McKee, and Skidmore (1993), have explored the determinants of a state's decision to adopt a lottery. (1) The optimal design of lottery games in terms of maximizing sales was studied by Quiggin (1991), Cook and Clotfelter (1993), and Garrett and Sobel (1999). Whether lottery ticket purchases are substitutes or complements for other consumer goods has been explored by Borg, Mason, and Shapiro (1993) and Kearney (2005). The revenue impact of cross-border lottery shopping has been studied by Garrett and Marsh (2002) and Tosun and Skidmore (2004). Similarly, Brown and Rork (2005) examined the strategic interaction between state lotteries using a model of tax competition. Finally, because states earmark net lottery revenues for programs such as education, several studies have explored whether lottery revenues increase spending on the target program (Borg and Mason, 1990; Spindler, 1995; Novarro, 2005).
The issue that has received the greatest attention is the tax incidence of lottery ticket expenditures for different income groups (Clotfelter and Cook, 1987, 1989; Scott and Garen, 1994; Hansen, 1995; Farrell, Morgenroth, and Walker 1999; Price and Novak, 1999; and Forrest, Gulley, and Simmons 2000). (2) Studies have used data of various levels of aggregation, such as individual survey data as well as aggregate data for zip codes, cities, counties, and states. (3) The majority of research has shown that state lotteries can be characterized as regressive taxation, implying a decreasing tax burden in relative terms as incomes rises. (4) Not surprisingly, this tax regressivity is raised as an objection to state lotteries, especially in light of the revenue maximization objective of state lottery agencies.
In most lottery demand studies a single income elasticity of demand is estimated from the sample of data, thus providing a static look at the tax incidence of lottery tickets. The results from previous research allow a comparison of a single income elasticity estimate from one study (state, county, city, or zip code at a point in time) with another study (another state, county, city, or zip code at a point in time), but little evidence exists on how the income elasticity of demand for a specific state's lottery product has changed over time. As discussed in the next section, rising consumer income, the introduction of new games, and other changes in the gambling landscape suggest that the tax incidence of lottery expenditures has not remained constant over time.
This paper provides evidence on the dynamic nature of the income elasticity of demand for lottery tickets. Using a panel of annual county-level data for three states that have had a lottery for 17 or more years, we estimate annual income elasticities of demand for each state. This provides a sufficient time series to track how the income elasticity of demand for lottery tickets and, thus, the lottery tax burden in each state has changed over the past several years. (5) We link the trends in the income elasticity of demand over time to possible business cycle effects, the introduction of new lottery products, and changes in consumer income. Our framework also allows us to explore whether growing competition for lottery products from casino gambling and neighboring state lottery games has changed the income elasticity of demand for lottery tickets.
Our results shed new light on the dynamic nature of lottery tax burdens and suggest that income elasticities estimated from a single year of data may accurately reflect the income elasticity of demand only for a specific year. The degree of lottery regressivity or progressivity is time dependent. Our annual estimates also provide a picture of the long-term revenue prospects of the state lotteries studied here, an important issue for all lottery states and programs funded by lottery revenue. Our evidence suggests that the long-term revenue prospects of state lotteries are unfavorable compared to other sources of state tax revenue.
INCOME ELASTICITIES OVER TIME: THEORY
Numerous reasons exist that suggest the income elasticity of demand for lottery tickets is likely to change over time. The most general is that consumer income has generally increased over time. A consistent empirical finding is that as income rises, individuals spend a smaller share of their budgets on lottery tickets (i.e., the income elasticity is less than one). Thus, as incomes tend to rise over time, the income elasticity of lottery demand should decline.
In addition to changes in income, the competitive environment in which lotteries operate is ever-changing. Relative to the mid 1980s, today's environment includes different advertising campaigns, new lottery games, more gaming alternatives (e.g., casino gaming and lotteries in neighboring states), and increased information about games. Precisely how these changes have affected the income elasticity of demand is uncertain.
Consider how advertising expenditures might affect the income elasticity of lottery demand. Increased lottery-related advertising should lead to increased spending on lottery tickets; however, the effect of increased advertising expenditures on the income elasticity of lottery demand depends on how the spending patterns of individuals change. (6) The income elasticity could increase, decrease, or remain unchanged. (7) Many advertising campaigns are targeted to certain groups, so specific advertising campaigns are undertaken with an expectation of affecting groups differentially. (8) A campaign targeting low-income players should tend to reduce the income elasticity of lottery demand, while a campaign targeting high-income players should tend to increase the income elasticity of lottery demand. A final observation is that an advertising campaign is often tied to the introduction of a new game, which itself might be targeted to appeal to a certain audience.
New games may also result in changing income elasticities of demand over time. Many states have added instant and online games that offer higher jackpots. For example, many states participate in multi-state online lottery games that generate much larger jackpots than could be offered from relying on lottery sales in a single state. In addition, starting in the 1990s many states began to offer $2, $5, or even $10 and $20 instant lottery games that offer much higher jackpots ($1 or $2 million) and payout rates (above 60 percent) than the traditional $1 instant game (payout rates averaging 50 percent). (9) Research by Mikesell (1989) and Oster (2004) suggests that large-jackpot online lottery games may attract a wealthier player than lower-jackpot instant games and, thus, may increase the income elasticity of demand. (10)
Furthermore, the magnitude of the change in the income elasticity of lottery demand likely differs for instant lottery games and online lottery games. Instant lottery games typically offer prizes ranging from about $100,000 to $250,000, with some offering a top-prize of $1 or $2 million. (11) Although these amounts have increased over time, the top-prize amounts for online lotto games have increased much more--top prizes in the 1980s and early 1990s were typically several million dollars, whereas today top prizes reach hundreds of millions of dollars.
Income elasticities of lottery demand may also change over time due to the introduction of lotteries in neighboring states and by the introduction of casino gaming. The start of a lottery by a neighboring state is likely to have larger effects on lottery spending in a state's counties that border the state than on other (i.e., internal) counties. A similar comment pertains to casino gaming in that, for a specific state, spending on lotteries in counties with casinos is likely to be affected more than counties without casinos. The key to any change in the income elasticity of lottery demand is whether spending in the border (or casino) counties is affected differently. For example, if lottery spending in each border county is affected identically in percentage terms by the introduction of a lottery in a border state, then the income elasticity of lottery demand will be unchanged for both the border counties and the entire state. However, even when the individual border counties and the casino counties are affected differently relative to each other, it is uncertain how the income elasticity for the entire state will change. (12) If the income elasticity of the border or casino counties increases, the state's income elasticity may increase, decrease, or remain unchanged.
In sum, changes in the income elasticity of demand for lottery tickets likely depend upon various factors, such as income growth, lottery advertising, the introduction of new games, whether a neighboring state has a lottery or has introduced a lottery, and the extent of casino gaming. Because these factors differ across states, it seems reasonable to assume that the levels and trends of the income elasticity of lottery demand also differ across states.
INCOME ELASTICITIES OVER TIME: PREVIOUS EVIDENCE
Few studies have explored how the income elasticity and tax regressivity of lottery sales have changed over time. For those few studies, the time period covered has been too short to allow definitive conclusions about changes over time. (13)
Mikesell (1989) used annual data for a subset of counties in Illinois from 1985-1987. (14) Three major findings emerged. First, the income elasticity for instant lottery games tended to be less than for online games. Second, all income elasticities did not differ statistically from one, so there was no evidence of tax regressivity. Third, the income elasticity for total lottery sales always exceeded one and increased from 1985 through 1987.
Jackson (1994) examined lottery sales, in total and for three separate games, for 1983 and 1990 in cities with 15,000 or more residents in Massachusetts. He found that the estimated income elasticities declined significantly for total sales and for each of the three games. The declines in each case were so large that the lottery changed from being a highly progressive tax in 1983 to a regressive tax in 1990. In addition, the income elasticities for instant games were less than those for online games.
A final study using multiple years and county-level data for five states was conducted by Hansen et al. (2000). They found that the lottery tax for the five states tended to be regressive, but no consistent finding emerged with respect to changes in the income elasticity of lottery demand. The...
|
