Taxing the Poor
Friday, June 22, 2007
by the National Center for Policy Analysis Task Force on Taxing the Poor
Table of Contents
Section I: Excise Taxes and Tobacco
Tobacco taxes have become a way for policymakers to increase state revenues without encountering the voter backlash that would come with an overt hike in tax rates. Besides, tobacco taxes are often perceived as a win-win: The tax is only paid by those who engage in socially undesirable behavior and, if smokers quit to avoid the tax, they - and society - are better off. But policymakers should also be concerned with the economic well-being of their lower-income constituents. One must question the fairness of hiking taxes that are known to disproportionately burden poor families.
“Tobacco taxes are more regressive than almost any other tax.”
Like many other popular ways of raising government revenue, excise taxes on tobacco are regressive. In fact, tobacco taxes are more regressive than almost any other tax. People with lower incomes spend a larger share of income on tobacco products than the wealthy. If rich and poor people consume the same amounts of a good, the poor necessarily spend a greater portion of their income on that good. But in the case of tobacco, smokers are much more likely to have lower incomes than nonsmokers, making the regressivity problem even worse.
History of Tobacco Taxes
Excise taxes are levied by both the federal and state governments - and in some cases, by municipal governments, too. While the United States has flirted with taxes on tobacco products since its founding, the federal government began vigorously taxing tobacco in 1862 to meet the fiscal demands of the Civil War.
Iowa imposed the first state tax on tobacco in 1921. Soon after, a number of other state legislatures, recognizing that tobacco taxes carried minimal political costs, imposed their own taxes:
- By 1950, 40 states and the District of Columbia charged between 1 cent and 8 cents for a package of 20 cigarettes.
- Between 1950 and 1962, 40 of the 47 states taxing cigarettes raised their rates at least once.
- In 1969, North Carolina became the last state to begin collecting excise taxes on cigarettes and other tobacco products.
Figure I-1 depicts the recent history of state and federal tobacco excise taxes. Real (inflation-adjusted) taxes per pack rose about 10 percent from 1955 through the early 1970s, then declined steadily for a decade before a federal excise tax increase in 1983. 1 They remained relatively constant in real terms until federal excise taxes were increased again in 1991 and 1993. Since then, excise taxes have continued to rise, largely due to state actions.
“Taxes per pack of cigarettes have been increasing since 1983.”
Tobacco Taxes Today . The current federal excise tax on cigarettes is 39 cents per pack; other tobacco products are subject to similar levies. Federal tobacco taxes raised about $7.8 billion in 2005, or about 0.36 percent of total federal revenues. 2 State governments raised about $11.4 billion from tobacco taxes in 2005. 3 New York and California each collected more than $1 billion, while Michigan and Pennsylvania each collected almost $900 million. 4 State levies on cigarettes are shown in Table I-1 . The wide variation is striking:
- States in the Northeast have the highest rates - $2.46 per pack in Rhode Island and $2.40 per pack in New Jersey.
- States with close ties to the tobacco industry have the lowest rates; Kentucky and North Carolina charge 3 cents and 5 cents per pack, respectively.
- The median rate is about 70 cents per pack.
Since the mid-1990s, states have also collected billions of dollars from tobacco companies in lawsuit settlements. As the sidebar shows, the costs are largely borne by consumers via higher prices. [See the sidebar, "The Tobacco Tort Tax."] In essence, the costs imposed by lawsuit settlement payments can be viewed as a tax, although it is difficult to determine the cost per pack, since recent price increases appear to exceed the settlement costs.
Sidebar: The Tabacco Tort Tax
The Regressive Nature of Tobacco Taxes
Most tax studies assume tobacco taxes will be passed on to consumers in the form of higher prices. Are they right? This assumption is consistent with standard microeconomic theory suggesting that any event that drives up the costs of all producers in a competitive market will, after a period of adjustment, be completely reflected in the price of the good. 5 In markets that are not perfectly competitive, the impact of a price increase will depend on the level of market power enjoyed by each firm, the market demand and the nature of the interaction between firms. For example, if one producer monopolizes the industry, and the demand for the product remains constant, an excise tax might actually raise prices by more than the amount of the tax. In an oligopolistic industry, such as tobacco, a few dominant firms might coordinate price increases around any changes in taxes.
The flurry of tobacco tax increases imposed by local, state and federal governments over the past 30 years provides a natural experiment for testing the effect of taxes on prices. A number of researchers have done so and found that almost all (and sometimes more) of a tax increase is passed on in the form of higher prices. 6
“The poor, less-educated and less-skilled pay a greater share of their incomes for tobacco products than the wealthy, highly educated and skilled.”
Who Smokes? If both the rich and the poor consume the same amounts of a good, the poor necessarily spend a greater portion of their income on that good. But in the case of tobacco, lower-income individuals are more likely to be smokers than those with higher incomes. As a result, no matter how tobacco taxes are measured, lower earners spend more on tobacco products and necessarily bear a heavier burden. Consider:
- Almost 29 percent of adults with incomes of less than $15,000 are smokers; by contrast, only 17.2 percent of people with incomes higher than $50,000 smoke. [See Figure I-2.]
- Nearly 40 percent of people with only a GED [General Educational Development] certificate are smokers; just 11.7 percent of people with a college degree - and 8 percent of people with a graduate degree - are smokers. 7
Lower Earners Pay Relatively More Tobacco Taxes . Since low earners are more likely to consume tobacco products, the poor, the less-educated and the less-skilled pay a greater share of their incomes for tobacco products than the wealthy, the highly educated and the skilled:
- People in the bottom 20 percent of income earners, as a group, spend 2.33 percent of their income on tobacco products, more than 10 times the percentage of income paid by the highest earners. [See Table I-2 .]
- High school dropouts spend about 1.2 percent of income on tobacco products, versus 0.12 percent for those with a postgraduate degree.
- Laborers and construction workers spend 1.0 percent and 1.1 percent of their incomes on tobacco products, respectively, versus just 0.24 percent for managers and professionals. 8
Less-skilled and less-educated individuals also spend more on average on tobacco products in any given year than do higher earners. (This highlights the unique nature of tobacco products: For almost any other good - housing or entertainment, for example - higher earners spend more than lower earners.)
- On the average, laborers and construction workers spend far more per year on tobacco products ($445 and $557, respectively) than managers and professionals ($211).
- High school graduates spend much more each year ($380) than people with professional degrees ($113). 9
These remarkably different patterns of consumption occur despite a large earnings gap: Professionals earn three times the wage of high school dropouts and twice the wage of high school graduates, on average. And managers earn almost twice the wage of laborers and two-thirds more than construction workers.
“Tobacco spending decreases significantly as income, education and career status increases.”
Lower-Earning Smokers Bear a Significantly Higher Burden . The difference becomes even more pronounced when one considers only households that report tobacco expenditures. Examining actual smokers provides a more meaningful comparison of the effects of excise taxes on different income groups. Unfortunately, there is little data on how much households reporting tobacco expenditures spend on tobacco products by income level but, as noted, education and occupation provide a rough proxy for income. According to the U.S. Bureau of Labor Statistics: 10
- High school dropouts who smoke spend an average of $1,301 on tobacco products each year; high school graduates who smoke spend $1,453.
- By contrast, smokers with professional degrees spend $1,248.
- Therefore, professionals who smoke spend only 1.27 percent of income on tobacco products, compared to 4.47 percent of income for high school dropouts who smoke. 11
- Similarly, laborers who smoke spend $1,382 per year (3.07 percent of income), and construction workers who smoke spend $1,464 (2.84 percent of income).
- By contrast, managers and professionals who smoke spend $1,303 on tobacco products - only 1.5 percent of income.
“The poor spend four times as much of their total annual consumption on tobacco as the wealthy.”
Lifetime Income and Regressivity . Any study of tax regressivity - in fact, studies of any sort of inequality - must deal with an obvious but potentially troublesome problem: incomes change. For example, a 25-year-old law student at the top of her class at a premier law school is probably below the official poverty level. Yet no sensible person would describe her as poor, since her income will rise dramatically after graduation. Similarly, a 50-year-old entrepreneur who just sold a business may appear to have an extremely high income, but averaged over the years it took to build the business, the seeming bonanza looks rather modest.
Our hypothetical law student and entrepreneur may have very similar lifetime incomes, even though in the current year one appears to be poor and the other wealthy. If they were both smokers, both would pay about the same proportion of their lifetime income in taxes. Because incomes change over workers' lifetimes, these taxes may appear to be more regressive than is, in fact, the case. Therefore, an examination of the lifetime incidence of excise taxes is useful.
If households base current spending on their expected lifetime incomes, then consumption provides a more accurate measure of lifetime resources than annual incomes. In order to determine the lifetime incidence of excise taxes, economist James Poterba compared annual expenditures on taxed goods to total consumption rather than to pretax income. 12 Poterba explored the different effects annual and lifetime incomes have on the regressivity of excise taxes on gasoline, alcohol and tobacco. As a share of expected lifetime income, expenditures on gasoline and alcohol are still regressive, but generally more equal than as a share of annual income. But tobacco is an exception. Poterba says, "For tobacco, however, even using the consumption metric, the excise tax appears regressive." 13
Repeating the previous comparison of the tobacco purchases of rich and poor, but using total expenditures instead of total income [shown in the far right column of Table I-2 ]:
- Tobacco spending as a percentage of total annual consumption is four times greater for the lowest earners than for the highest earners, versus 10 times greater when calculated as a percentage of total annual income.
- Even when calculated as a percentage of total expenditures, tobacco taxes are still regressive, whether considered by income, education or occupation.
The poor are more likely to smoke, more likely to spend a greater share of income on tobacco products and, in some cases, spend more on tobacco products in real dollar terms than their wealthier counterparts. Thus, proportionately, the poor are more likely to bear a heavier burden of tobacco taxes. By these measures, tobacco taxes are regressive.
Excise Tax Increases and Behavioral Responses
The previous comparisons are based on simple definitions of regressivity - that is, taxes paid relative to income and consumption. However, if different income groups respond in different ways - or to different degrees - to a particular tax, their behavior may affect the regressivity of the tax.
Rising Costs Do Not Affect All Consumers the Same Way . The evidence suggests that the poor are more responsive to changes in prices than the rich, meaning the poor are more likely to quit or reduce consumption of tobacco products than the wealthy when prices rise. Thus, simple comparisons that assume an equal response to high cigarette taxes among all income groups may tend to overstate the regressivity.
Economists generally measure responsiveness to price using elasticity of demand: The percentage change in tobacco products divided by the percentage change in price.
- An elasticity of exactly one, in absolute value, means that a percentage change in price results in an equal percentage change in quantity demanded.
- If elasticity is greater than one, in absolute value, the price increase will result in an even greater percentage decline in the quantity demanded and, consequently, less spending.
- If elasticity is less than one, in absolute value, the price increase will not be matched by an equal percentage decline in the quantity demanded, and spending will rise.
Purchases of addictive goods like cigarettes are considered to be fairly insensitive to price increases. According to a review of the literature by Dahlia Remler, elasticity of demand for tobacco products is between -0.3 and -0.5, indicating these purchases are generally "insensitive to price, but certainly not completely insensitive." 14 In other words, for a typical smoker, a 10 percent price increase leads to a 3 percent to 5 percent decrease in consumption. (Stated another way, if cigarette prices increase, a typical smoker's spending on the product will rise since his decrease in consumption is less than the increase in price.)
However, there is ample empirical evidence that smokers as a group are responsive to price increases. National Bureau of Economic Research Associate Michael Grossman finds that changes in price explain a "good deal" about the consumption rates of addictive substances like tobacco products, particularly among teenagers - a result consistent with the growing body of evidence. According to Grossman, "The dramatic increase in the price of cigarettes since 1997 explains almost all of the 12-percentage-point reduction in the cigarette participation rate [among teenagers] since that year." 15
“Higher tobacco taxes make low-income smokers worse off because they can no longer afford something they want.”
Differential Effects among Rich and Poor Smokers. One of the few studies measuring the difference between the responsiveness of lower- and higher-income smokers to price increases was conducted by Matthew C. Farrelly and his colleagues for the U.S. Centers for Disease Control and Prevention. The study, which examined 14 years of data from the National Health Interview Survey, suggests lower-income smokers (with an elasticity of -0.29) are more likely to reduce their tobacco expenditures when faced with higher prices than smokers with higher incomes (elasticity of -0.17). 16
Farrelly conducted a similar study several years later, which also concluded that lower-income populations were more likely to reduce or quit smoking than their wealthier counterparts, but by a wider margin. It found lower-income smokers (with an elasticity of -0.43) are more than four times as likely to reduce their tobacco expenditures as smokers with higher incomes (elasticity of 0.00 to -0.10). 17
The result: Because higher tobacco taxes encourage the poor to reduce expenditures to a greater degree than the wealthy, the total tax burden shifts to higher-income groups. 18 Unfortunately, lower-income smokers as a group still pay a larger share of their income in excise taxes. Further, regardless of the extent to which lower-income smokers kick the habit, most would agree the lower-income group is worse off because they can no longer afford something they want.
Measures of Well-Being and Paternalism 19
Traditional measures of regressivity may be lacking because they exclude measures of consumer welfare, or the satisfaction derived from consuming a product. Economists base welfare calculations on consumers' willingness to pay for a good, which is assumed to reflect its value to them.
According to this argument, excise taxes on tobacco hurt smokers - particularly those with lower incomes - in two ways. First, those who don't quit the habit pay higher prices and are worse off. Second, those who stop consuming are worse off because they no longer get the value of what they wanted to consume. If welfare regressivity is a legitimate consideration when examining tobacco taxes, "externalities" - the costs smokers impose on nonsmokers - should also be included. See the sidebar, "Do Smokers Pay Their Own Way?" for a brief discussion of externalities. As the sidebar explains, the external costs are almost certainly much lower than the current level of taxes.
“Smokeless tobacco is less harmful than cigarettes, but some states tax it more.”
Are excise taxes defensible on some grounds, even if they tax the poor? Consider the differential taxation of smokeless tobacco and cigarettes, which is fully explored in the sidebar. Although smokeless tobacco is not safe, studies suggest it is safer than smoking cigarettes. If the true purpose of excise taxes on tobacco products is to recoup the external costs to society, states should levy lower taxes on smokeless products; however, some states (Massachusetts, Texas Minnesota, Idaho, Oklahoma and others) do just the opposite. Since the less harmful product faces the highest tax in many states, the motive is not to recover the external cost.
Some advocates of high tobacco taxes acknowledge the regressivity of excise taxes but focus on other ways to make the overall tax system more progressive. 20 Other advocates of higher cigarette taxes suggest tobacco taxes are actually progressive when welfare and health effects for the poor are included.
In fact, a relatively new field of behavioral economics says traditional economic welfare measures are not appropriate. These scholars - among whom Jonathan Gruber is a leading voice - expand on the theory of externalities and focus on "internalities," suggesting smokers' willingness to pay for tobacco products does not properly account for the harm smokers do to themselves. This assumes all smokers want to quit and seek a "commitment device," or an external catalyst to help break their addiction. 21 Due to the greater responsiveness of the poor to price increases, once the "self-control benefits" of taxes are accounted for, says Gruber, tobacco taxes reduce regressivity and may even be progressive in some cases. 22
“The external costs to society of smoking are almost certainly much lower than the taxes smokers pay.”
There are drawbacks to Gruber's approach. First, it may simply drive smokers to substitute illegal, untaxed purchases for those that are legal and taxed at high rates. Illegal cigarette trafficking has become a multi-billion-dollar-a-year, worldwide trade. It sprang up in response to the large difference between wholesale and consumer prices, which result from high tobacco taxes around the world, according to a study conducted by the Government Accountability Office. 23
Sidebar: Do Smokers Pay Their Own Way?
Second, it requires researchers to make many assumptions about the behaviors and preferences of their subjects, the validity of which is highly uncertain. Further, to put it bluntly, this view holds that smokers are ignorant of their own best interests and that government intervention can prevent them from harming themselves.
“The poor are more likely than the wealthy to give up smoking in response to a price increase.”
Eight out of 10 smokers say they want to quit smoking. 24 Some do quit, but many don't. Why not? It may simply be that in answering such survey questions, respondents use the word "want" to indicate some level of desire and not an intended course of action. If, for example, researchers asked men between the ages of 18 and 25 whether they wanted a Corvette, many would answer yes. But only of few of those who say they want the car actually buy one. To experience the pleasure of owning a Corvette requires the sacrifice of many other pleasures. To capture the benefits of not smoking requires the sacrifice of the pleasure of smoking. If smokers are acting rationally - in the sense that the benefits of smoking are greater than the costs - then taxes that force people to quit or reduce smoking make them worse off. And by extension, therefore, taxes that disproportionately impact poor smokers are regressive.
Several economists have suggested that tobacco use is different than owning a Corvette. Because smoking is not only a learned behavior but actually addictive, these analysts claim that smoking is irrational and so interventions forcing people to quit smoking can actually leave them better off. Someone who buys a Corvette this year should know that he or she will spend more on gasoline and car insurance in the future. Similarly, someone who begins smoking this year should know that he or she is likely to continue the habit in the future.
Expressed this way, the argument is seen as Big Brother paternalism dressed up in the esoteric language of economics. And whether the government can or should protect us from our own folly is an important and unresolved matter that cannot be settled by scientific analysis.
“Lower-income people who don't quit will bear a heavier burden than the wealthy.”
The poor, the less-educated and the less-skilled are more likely to smoke, more likely to spend a greater share of income on tobacco products and, in some cases, spend more on tobacco products in real dollar terms than their wealthier counterparts. Thus, proportionately, the poor are more likely to bear a heavier burden of tobacco taxes. By these measures, tobacco taxes are regressive. At the same time, the poor are more likely to quit smoking - or reduce their tobacco expenditures - in response to higher prices.
Although some suggest that higher tobacco taxes will make the poor better off because of the health effects of quitting smoking, by conventional economic measures, taxing tobacco means taxing the poor. Moreover, if prices rise, lower-income people who don't quit will pay higher prices.
By any measure, tobacco taxes place a heavier burden on lower-income families. As a result, excise taxes on tobacco products are regressive. Policymakers who are considering raising tobacco taxes at the federal, state or local level should consider the disproportionate burden their lower-income constituents will bear.
References: Table I-2
1 Jonathan Gruber, "Tobacco at the Crossroads: The Past and Future of Smoking Regulation in the United States," Journal of Economic Perspectives , Vol. 15, No. 2, spring 2001, pages 193-212.
2 Congressional Budget Office, Budget Options (Washington, D.C.: Congress of the United States, February 2005), page 330.
3 U.S. Census Bureau, Census of Governments, State Government Tax Collections Summary Tables, March 2005. Available at http://www.census.gov/govs/www/statetax.html.
5 To state the point more carefully, in a perfectly competitive industry where all firms have similar costs, the long-run equilibrium price is defined by the average costs (including all opportunity costs) of a firm producing at minimum efficient scale and hence minimum average costs. Since an excise tax will raise minimum average costs by the amount of the tax, the price will include the tax.
6 A study by William N. Evans et al. presents a compelling analysis using data from a 12-year period from all states and the District of Columbia, finding "that excise taxes are passed completely on to consumers in the form of higher retail prices." See William N. Evans, Jeanne S. Ringel and Diana Stech, "Tobacco Taxes and Public Policy to Discourage Smoking," Tax Policy and the Economy , Vol. 13, 1999. For more information, see Jeffrey E. Harris, "On the Fairness of Cigarette Excise Taxation," in The Cigarette Excise Tax (Cambridge, Mass.: Harvard University, 1985), pages 106-111; Daniel A. Sumner, "Measurement of Monopoly Behavior: An Application to the Cigarette Industry," Journal of Political Economy , 1981; Daniel Sullivan, "Testing Hypotheses about Firm Behavior in the Cigarette Industry," Journal of Political Economy , 1985; Nicholas H. Stern, "The Effects of Taxation, Price Control and Government Contracts in Oligopoly and Monopolistic Competition," Journal of Public Economics , Vol. 32, No. 2, 1987, pages 133-58; Paul G. Barnett, Theodore E. Keeler and Teh-wei Hu, "Oligopoly Structure and the Incidence of Cigarette Excise Taxes," Journal of Public Economics , Vol. 57, No. 3, 1995, pages 457-70; and Theodore E. Keeler et al., "Do Cigarette Producers Price-Discriminate by State? An Empirical Analysis of Local Cigarette Pricing and Taxation," Journal of Health Economics , Vol. 15, No. 4, 1996, pages 499-512.
7 Income data from William N. Evans, Jeanne S. Ringel and Diane Stech, "Tobacco Taxes and Public Policy to Discourage Smoking." Education data from the Centers for Disease Control and Prevention's "National Health Interview Survey, 2004."
8 "Consumer Expenditures in 2004," U.S. Department of Labor, Bureau of Labor Statistics, Report No. 992, April 2006, Tables 1, 12 and 13. Available at http://www.bls.gov/cex/csxann04.pdf.
10 Mark Vendemia, "Tobacco Expenditures by Education, Occupation and Age," in "Consumer Expenditure Survey Anthology, 2005," U.S. Department of Labor, Bureau of Labor Statistics, Report No. 981, April 2005, page 51. Available at http://www.bls.gov/cex/csxanthol05.pdf.
11 Incomes for the education and occupation groups are presented in Table I-2.
12 James Poterba, "Lifetime Incidence and the Distributional Burden of Excise Taxes," American Economic Review , Vol. 79, 1989, pages 325-30.
14 Dahlia K. Remler, "Poor Smokers, Poor Quitters and Cigarette Tax Regressivity," American Journal of Public Health , Vol. 94, No. 2, February 2004, pages 225-29. For a more thorough treatment of smokers' responsiveness to high cigarette taxes, see Michael Grossman, "Individual Behaviors and Substance Abuse: The Role of Price," National Bureau of Economic Research, Working Paper No. 10948, December 2004.
15 Matt Nesvisky, "Smoking, Drinking and Drug Use Respond to Price Changes," National Bureau of Economic Research, NBER Digest, May 2005; based on Michael Grossman, "Individual Behaviors and Substance Abuse: The Role of Price." Accessed August 17, 2006. Available at http://www.nber.org/digest/may05/w10948.html. Of course, as Grossman acknowledges, price is not the only causal factor. Developments in antismoking campaigns, the diffusion of knowledge about the harmful effects of cigarette smoking, minimum legal ages, warning labels and the price of nicotine replacement products like gum and patches also play a role. After all, the proportion of adults who smoke declined steadily from 1966 to 1990, despite the fact that prices rose, fell and rose again during that period. For a more detailed discussion, see Jeffrey E. Harris, "A Working Model for Predicting the Consumption and Revenue Impacts of Large Increases in the U.S. Federal Cigarette Excise Tax," National Bureau of Economic Research, Working Paper No. 4803, July 1994.
16 Matthew C. Farrelly and Jeremy W. Bray, "Response to Increases in Cigarette Prices by Race/Ethnicity, Income and Age Groups - United States, 1976-1993," Morbidity and Mortality Weekly Report , Vol. 47, No. 29, July 1998, pages 605-9.
17 Matthew C. Farrelly et al., "Response by Adults to Increases in Cigarette Prices by Sociodemographic Characteristics," Southern Economic Journal , Vol. 68, No. 1, 2001, pages 156-65.
18 Since the poor are more likely to reduce tobacco expenditures than the rich, it reasonable to assume rising prices will shift the burden of taxation to the rich. However, it should be noted that over the past several decades, smoking has become "very socioeconomically concentrated," even as cigarette prices have increased. This could happen for reasons other than price elasticity, but whatever the cause, it has more than offset the effect of a higher price elasticity among the poor. See Jonathan Gruber, "Tobacco at the Crossroads."
19 This section draws on concepts explored in Dalhia Remler, "Poor Smokers, Poor Quitters and Cigarette Tax Regressivity."
20 One such advocate is Edgar Browning, professor at Texas A&M University. Browning says the poor receive large portions of their incomes from government transfer programs, which are funded directly or indirectly from excise taxes and other government revenues. Since the rich bear most of the tax burden and most of the benefits flow to lower-earners, the entire system as a whole is progressive. Edgar K. Browning and William R. Johnson, "The Distribution of the Tax Burden," American Enterprise Institute, 1979. Also see Edgar K. Browning, "Tax Incidence, Indirect Taxes and Transfers," National Tax Journal , Vol. 38, December 1985, pages 525-34.
21 Jonathan Gruber and Botond Koszegi, "A Theory of Government Regulation of Addictive Bads: Optimal Tax Levels and Tax Incidence for Cigarette Excise Taxation," National Bureau of Economic Research, Working Paper No. 8777, February 2002.
23 "Cigarette Smuggling: Federal Law Enforcement Efforts and Seizures Increasing," U.S. General Accounting Office, Report No. GAO-04-641, May 2004. Available at http://www.gao.gov/new.items/d04641.pdf.
24 David Burns, "Reducing Tobacco Use: What Works in the Population?" Journal of Dental Education , Vol. 66, No. 9, 2002, pages 1,051-60.