How Effective Is the Minimum Wage at Supporting the Poor? a

Transcription

1 How Effective Is te Minimum Wage at Supporting te Poor? a Tomas MaCurdy b Stanford University Revised: February 2014 Abstract Te efficacy of minimum wage policies as an antipoverty initiative depends on wic families benefit from te increased earnings attributable to minimum wages and wic families pay for tese iger earnings. Proponents of tese policies contend tat employment impacts experienced by low-wage workers are negligible and, terefore, tese workers do not pay. Instead proponents typically suggest tat consumers pay for te iger labor costs troug imperceptible increases in te prices of goods and services produced by low-wage labor. Adopting tis best-case scenario from minimum-wage advocates, tis study projects te consequences of te increase in te national minimum wage instituted in 1996 on te redistribution of resources among ric and poor families. Under tis scenario, te minimum wage increase acts like a sales tax in its effect on consumer prices, a tax tat is even more regressive tan a typical state sales tax. Wit te proceeds of tis national sales tax collected to fund benefits, te 1996 increase in te minimum wage distributed tese bulk of tese benefits to one in four families nearly evenly across te income distribution. Far more poor families suffered reductions in resources tan tose wo gained. As many ric families gained as poor families. Tese income transfer properties of te minimum wage document its considerable inefficiency as an antipoverty policy. a Aspects of te arguments and approac in tis study ave appeared in several non-peer reviewed reports and working papers written by te autor (and different coautors) since te late 1990s. Tese reports/papers greatly benefited from discussion, comments and expert researc assistance from Frank McIntrye, Peggy O Brien-Strain and Selen Opcin. For tis updated paper and newly produced empirical results, te autor gratefully acknowledges many useful contributions from Kevin Mumford. b Professor, Department of Economics, and Senior Fellow, Te Hoover Institution, Stanford University, Stanford, CA tmac@stanford.edu

2 1. Introduction Te widespread popularity of raising te minimum wage draws eavily on its appeal as an antipoverty policy, wic relies on two beliefs: first, raising te minimum wage will increase te incomes of poor families; and second, te minimum wage imposes little or no public or social costs. Indeed, in 2006 a group of more tan 650 economists signed a widely distributed statement issued by te Economic Policy Institute expressing tese sentiments in support of legislation calling for a 40% increase in te federal minimum wage. Tis support along wit broad acceptance of tese beliefs encouraged policymakers in Wasington DC to raise te minimum wage from $5.15 in 2007 to $7.25 in Te policy debate over te minimum wage principally revolves around its effectiveness as an antipoverty program. A popular image used by bot sides of te debate consists of families wit breadwinners wo earn low wages to support teir cildren. Policies tat raise te wages of tese workers increase teir earnings and contribute to teir escaping poverty. As a counterbalance to tis impact, opponent of te minimum wage argue tat wage regulation causes some low-wage workers to lose teir jobs and tey will suffer income drops. Te issue, ten, becomes a tradeoff; some low-income breadwinners will gain and oters will lose. Promoters of te minimum wage retort tat employment losses are quite small and, consequently, te workers wo gain far exceed tose wo lose. In addition to potential adverse employment effects, opponents of minimum wages furter counter te belief tat te minimum wage assists poor families by documenting tat many minimum-wage workers are not breadwinners of low-income families. Tey are, instead, often teenagers, single eads of ouseold wit no cildren, or not even members of low-income families. Promoters of te minimum wage admit tat some of tese groups may also benefit from te wage increase, but since few workers lose jobs, tey contend tat te minimum wage still benefits low-income families wit cildren. Te notion tat te minimum wage can be increased wit little or no economic cost underlies many advocates assessments of te effectiveness of te minimum wage in its antipoverty role. Most economists agree tat imposing wage controls on labor will not raise total income in an economy; indeed, elementary economics dictates tat suc market distortions lead to reduced total income implying fewer overall benefits tan costs. If, owever, one presumes tat employment losses do not occur and total income does not fall, ten te minimum wage debate becomes a disagreement over ow it redistributes income. Te efficacy of a minimum wage ike as an antipoverty program depends on wo benefits from te increase in earnings and wo pays for tese iger earnings. Wereas a number of studies ave documented wo benefits, wo pays is far less obvious. But someone must pay for te iger earnings received by te low-wage workers. At te most simplistic level, te employer pays for te increase. However, businesses don't actually 1

3 pay, for tey are merely conduits for transactions among individuals. Businesses ave tree possible responses to te iger labor costs imposed by te minimum wage. First, tey can reduce employment or adjust oter aspects of te employment relationsip (e.g., less fringes or training opportunities), in wic case some low-wage workers pay temselves troug loss of teir jobs or by receiving less non-salary benefits; second, firms can lose profits, in wic case owners pay; and, tird, employers can increase prices, werein consumers pay. Of tese tree sources, entertaining tat low-wage workers bear any cost of te minimum wage as been largely dismissed by proponents in recent years based on several (albeit muc disputed) studies tat found little or no job loss following istorical increases in federal and state minimum wages. Wile te extra resources needed to cover iger labor costs could teoretically come out of profits, several factors suggest tat tis source is te least likely to bear costs. Capital and entrepreneursip are igly mobile and will eventually leave any industry tat does not yield a return comparable to tat earned elsewere. Tis means tat capital and entrepreneursip, and ence profits, will not bear any significant portion of a tax imposed on a particular factor of production. Stated differently, employers in low-wage industries are typically in igly competitive industries suc as restaurants and retail stores, and te only option for tese low profit margin industries becomes lowering exposure to low-wage labor or raising prices. Wit jobs presumed to be unaffected, tis leaves iger prices as te most likely candidate for covering minimum wage costs. In fact, supporters of minimum and living wage initiatives often admit tat sligt price increases pay for iger labor costs following minimum wage ikes. To evaluate, ten, te redistributive effects of te minimum wage adopting te view implicitly eld by its advocates, tis study examines te antipoverty effectiveness of tis policy presuming tat firms raise prices to cover te full amount of teir iger labor costs induced by te rise in wages. In particular, te analysis simulates te economy taking into account bot wo benefits and wo pays for a minimum wage increase assuming tat its costs are all passed on solely in te form of iger consumer prices. Te families bearing te costs of tese iger prices are tose consumers wo purcase te goods and services produced wit minimum-wage labor. In actuality, most economists expect some of tese consumers would respond to te iger prices by purcasing less, but suc beaviors directly contradict te assertion of no employment effects since lower purcases mean tat fewer workers would be needed to satisfy demand. Consequently, to keep fait wit te view eld by proponents, te simulations carried out in tis study assume tat consumers do not alter teir purcases of te products and services produced by low-wage labor and tey bear te full cost of te minimum wage rise. Tis approac, ten, maintains te assumption of a steady level of employment, te best-case scenario asserted by minimum-wage proponents. 2

4 Altoug igly stylized and probably unrealistic, te following analysis demonstrates tat te minimum wage can ave unintended and unattractive distributional effects, even in te absence of te employment losses predicted by economic teory. To evaluate te distributional impacts of an increase in te minimum wage, tis study investigates te circumstances applicable in te 1990s wen te federal minimum wage increased from $4.25 in 1996 to $5.15 in To identify families supported by low-wage workers and to measure effects on teir earnings and income, tis analysis uses data from waves 1-3 of te 1996 Survey of Income and Program Participation (SIPP). To translate te iger earnings paid to low-wage workers into te costs of te goods and services produced by tem, tis study relies on national input-output tables constructed by te Minnesota Impact Analysis for Planning (IMPLAN) Group, matced to a time period comparable wit SIPP s. To ascertain wic families purcase te goods and services produced by low-wage workers and ow muc more tey pay wen prices rise to pay for minimum wage increases, tis study uses data from te Consumer Expenditure Survey (CES), again matced to te same time period as SIPP s. Te contribution of tis study is not to estimate te distribution of benefits of te minimum wage, nor is it to estimate te effect on prices; bot of tese impacts ave already been done in te literature. Instead te goal of tis paper is to put te benefits and cost sides togeter to infer te net distributional impacts of te minimum wage on different categories of families and to translate tis impact into a format readily accessible to economists and policymakers. To provide an economic setting for evaluating te distributional measures presented ere, tis study develops a general equilibrium (GE) framework incorporating minimum wages. Tis model consists of a two-sector economy wit te two goods produced by tree factors of production: low-wage labor, igwage labor, and capital. A particular specification of tis GE model justifies te computations performed in te analysis, and entertaining alterations in its beavioral elements permits an assessment of ow results migt cange wit alternative economic assumptions. Te model proposed ere goes well beyond wat is currently available in te literature, wic essentially relies on a Heckscer-Olin approac wit fixed endowments (supplies) of labor and capital inputs. In contrast, te GE model formulated in tis study admits flexible elasticities for bot input supplies and for consumer demand, as well as a wide range of oter economic factors. Seven sections make up te remainder of tis paper. Section 2 reviews te economics literature on te responses available to employers to pay for te iger labor costs imposed by te minimum wage, and it 1 Tis increase was done in two steps: an increase from $4.25 to $4.75 on October 1, 1996 and ten to $5.15 on September 1, Adjusting for inflation, te $5.15 minimum wage in 1997 is wort about $7.00 in

5 relates tese survey findings to te simulation metod used in tis paper. Section 3 overviews te metodology and data used to carry out te simulations of minimum-wage impacts. Section 4 caracterizes wo benefits from an increase in te national minimum wage, and Section 5 describes wo pays for tis increase. Section 6 calculates te net distributional effects of a rise in te minimum wage. Section 7 discusses limitations of te analytical approac used ere witin a coerent GE model of te distributional impacts of te minimum wage. Finally, Section 8 summarizes te findings. 4

6 2. Paying for te Minimum Wage Tis section reviews te economics literature on ow employers respond to te iger labor costs imposed by te minimum wage and relates te findings from tis literature to te simulation metod used in tis paper. Te distributional effects of a minimum wage increase depend in part on wo pays te costs of te policy cange. Te literature as focused on tree possible responses (not mutually exclusive): first, employers could respond by reducing te ours of work or number of employees (workers pay); second, firms could increase prices (consumers pay); and/or tird, businesses could not respond at all wic would leave tem wit lower profits (owners pay). Te first tree subsections below discuss te economic reasoning and evidence for eac of tese responses, and te last subsection specifies te assumptions maintained in te following simulation analysis. 2.1 Reducing Employment Economics researc on te minimum wage as predominantly focused on te issue of employment losses. Tis focus draws on a fundamental tenet of economic teory: all else being equal, agents purcase less of a good as its price rises. According to tis teory, not only will employers reduce teir employment to mitigate costs associated wit a minimum wage ike, tey will also tend to reduce output and/or increase te utilization of oter factors of production. For eac potential employee, te firm decides weter aving additional ours will increase te firm's revenue sufficiently to justify tat worker's wage. For some firms, te extra revenue generated by te least productive workers becomes insufficient to justify teir wages, so employment falls. In tis scenario, low-wage workers bear part of te cost of an increase in te minimum wage troug reduction in employment and ours of work (also possibly troug reductions in forms of compensation oter tan earnings). 2 Te vast majority of te debate over te minimum wage revolves around measuring te rate at wic a rise in te minimum wage affects employment. Prior to te 1990s, economists widely eld te view tat minimum wage increases primarily adversely affect te employment of young workers under age 25. In teir survey of twenty-five time series studies of yout employment publised between 1970 and 1981, Brown, Gilroy, and Koen (1982) conclude tat a 10% increase in te minimum wage can be expected to reduce teenage employment by 1 to 3% according to existing empirical evidence; in teir review of a smaller number of cross-section studies, te estimated decrease in teenage employment ranged of zero to over 3% for a 10% increase in te minimum wage. Te accumulated researc of tis era generally maintains tat young adults beyond te 2 In addition to reducing fringe benefits and training, minimum wage employers can also presumably demand greater effort from te minimum wage workers wo remain employed. Given te limited fringe benefits and training in tese jobs, effort may well present a more important margin of adjustment. 5

7 teenage years experience notably smaller negative employment impacts tan teir teenage counterparts. Researc in te 1990s onward callenged tis conventional wisdom troug a series of studies tat exploited variation in state-specific minimum wages above te federal level as a primary source of data to measure impacts of minimum wage. Tis literature, comprised of more tan 100 papers written over te past two decades, as become known as te new minimum wage researc. Te most influential work in tis literature finds no negative employment effects, and some studies even suggests tat employment increases in reaction to minimum wage ikes. Card and Krueger's 1995 book Myt and Measurement compiles some of te most prominent work in tis area. Card and Krueger (1994) examine fast-food employment in New Jersey and Pennsylvania before and after te 1992 increase in New Jersey s minimum wage. Wit point estimates suggesting a positive employment effect, Card and Krueger conclude, we believe tat, on average, te employment effects of a minimum-wage increase are close to zero (p. 383). Oter studies discussed in Myt and Measurement, including Katz and Krueger (1992), Card (1992a), and Card (1992b), furter support tis conclusion. More recent studies by Zavodny (2000), Card and Krueger (2000), Dube, Naidu, and Reic (2007), Dube, Lester, and Reic (2010), and Allegretto, Dube, and Reic (2011) produce similar findings. As economic rationales for explaining teir empirical findings, tis line of researc predominately cites two caracterizations of labor markets: a monopsonistic labor market of te sort discussed by Stigler (1946), and bilateral searc models wit eterogeneous workers of te sort proposed in Lang and Kan (1998). Tis callenge of te conventional wisdom about minimum wage impacts as not gone unanswered in te literature. Several studies directly critique te approaces used to derive te new conclusions (e.g., Kim and Taylor (1995), Deere, Murpy and Welc (1995), Welc (1995), Neumark and Wascer (2000), and Burkauser, Couc, and Wittenburg (2000)). Oters studies confirm te consensus view of te 1980s and find negative employment effects primarily concentrated among younger workers (e.g., Currie and Fallick (1996), Williams and Mills (2001), Neumark (2001), Neumark and Wascer (2002), and Neumark, Scweitzer, and Waser (2004)). 3 Furter, te surveys of Brown (1999) and Neumark and Wascer (2007) point out tat muc of te empirical work in te new researc actually estimates small and negative employment responses to increases in minimum wages. Neverteless, te widely-eld view today in te economics profession maintains tat relativelymodest increases in te minimum wage exert negligible impacts on employment. In particular, according to a survey of senior faculty from te top researc universities in te US conducted by te Initiative on Global 3 Te book entitled Minimum Wages publised in 2010 by Neumark and Wascer summarizes te findings of tese studies and many oters. 6

8 Markets (IGM), only 40% (confidence weigted) believe tat raising te federal minimum wage would make it noticeably arder for low-skilled workers to find employment. 4 Advocates of te minimum wage often cite suc consensus wen arguing tat impacts on employment can be ignored. 2.2 Raising Prices A cost of te minimum wage commonly acknowledged by its advocates concerns its impacts on prices. Te labor demand curve, wic leads to te basic conclusions about employment effects, assumes tat product prices are eld constant. Tis is a reasonable assumption for firms tat compete wit oter firms tat are not affected by te minimum wage increase, suc as overseas or ig tec firms tat employ iger-wage workers. However, many of te industries tat employ minimum wage workers do not compete in suc markets. Tese include te types of service industries tat make up te largest sare of low-wage employers: eating and drinking places and retail trade. For tese industries, an increase in te minimum wage principally represents an industry-wide increase in costs. Terefore, prices for low-wage goods will rise. (Output could also fall, depending on te price sensitivity of consumers, but tis reaction is presumed not to occur to avoid te implications for reduction in employment.) In tis price-increase scenario, some of te burden of te minimum wage increase falls on te consumers of low-wage products. Altoug rigorous researc on te subject is somewat limited, a body of work as developed examining te impact of a minimum wage on prices. Te basic teoretical predictions were first noted by Stigler (1946) and ave been furter described by Hamermes (1993) and Aaronson and Frenc (2007). Lemos (2008) surveys te empirical literature in tis area and presents evidence supporting te claim tat prices rise as a result of minimum wage increases. Syntesizing te findings of nearly 30 studies, tis survey assesses a range of 0.4 to 0.04 for te estimated elasticity of te rise in prices induced by a minimum wage increase. One set of studies directly estimates price impacts (e.g., Wessels (1980), Card and Krueger (1995), Aaronson (2001), MacDonald and Aaronson (2006), Lemos (2006), and Aaronson, Frenc, and MacDonald (2008)). Considering several specific examples, Aaronson (1997) explores te effects of increasing te minimum wage on restaurant prices using a competitive market model. Using several data sources on restaurant prices in te United States and Canada, Aaronson s results suggest tat restaurant prices rise almost one-for-one wit increases in labor costs; a 1% increase in te minimum wage is associated wit an increase in restaurant prices of approximately 0.07% in bot countries. Moreover, e 4 More precisely, 40% agree tat raising te minimum wage would adversely impact te employment of low-wage workers; 38% disagreed; and 22% are uncertain. Only 16% don t favor indexing te minimum wage to inflation as a desirable antipoverty policy. See ttp:// 7

9 finds tat tese price adjustments are sort-run penomena concentrated in te quarters before and after te enactment of te minimum wage increase. Instead of concentrating on a single industry, Wilson (1998) looks at te macroeconomic price response attributable to a minimum wage increase. Using te Mark 11 U.S. Macro Model, an econometric model of te U.S. economy, e estimates tat increasing te federal minimum wage by $1.00 per our over te course of two years (1999 and 2000) yields an increase in prices of 0.2 percentage points in 1999 and 0.1 percentage points in Lastly, Card and Krueger (1995) include information on price effects. Based on a comparison of price growt in New Jersey and Pennsylvania after a minimum wage increase in New Jersey, Card and Krueger (1995, p.54) conclude tat "prices rose 4% faster as a result of te minimum-wage increase." In teir cross-state comparisons, te impacts on prices are imprecise estimated. Still, Card and Krueger surmise tat two different sources of data (city-specific CPIs and observations on amburger prices collected by te American Camber of Commerce Researc Association) indicate te same pattern of faster price increases in areas more affected by minimum wage increases. In fact, tey find tat te relationsip between iger wages and tese iger prices approximates te labor sare of product costs, a result consistent wit te teory tat te majority of te costs are being passed on in iger prices. Anoter set of studies indirectly estimate price impacts of minimum wages using input output models to trace wage increase on te inter-industry flow of goods and services to simulate impacts on employment, output and prices in te aggregate economy and various market sectors. Assuming full passtroug effect, no substitution effects, no employment effects and no spillover effects, Wolf and Nadiri (1981) used an input output model and data from te Current Population Survey to estimate te price effects attributable to te 1963, 1972 and 1979 minimum wage increases. Tey estimate tat a 10% 25% minimum wage increase raises prices by 0.3% 0.4%. Under similar assumptions, Lee and O Roark (1999) use an input-output model to estimate price effects in te food and food-service industries. Tey calculate tat a 50-cent minimum wage increase would raise consumer prices of food and kindred products by approximately 0.3%. Moreover, te same increase would raise prices by 0.9% in eating and drinking establisments, an industry wit a iger concentration of minimum wage workers and a larger sare of labor costs. Tey also consider te potential impacts of wage spillovers tat refer to increases in wages tat occur for tose earning sligtly more tan te minimum wage. Tis spillover leads to consumer prices increasing sligtly more, but never by more tan 1.5% in eating and drinking establisments and by 0.4% in food and kindred products. Not all empirical studies find evidence of rising prices in response to a minimum wage increase. Katz and Krueger (1992), Macin et al. (2003), and Draca et al. (2011) do not obtain statistically significant 8

10 impacts. But tis evidence is not compelling since te predicted impacts of minimum wage on prices are small and price data is igly variable and influenced by many factors. Wile te precise magnitude of te responsiveness of prices to minimum wages ikes is not firmly establised, te direction of te price response seems clear. Most economists and policy makers accept te view tat iger minimum wages translate into iger prices for te goods and services produced eiter directly or intermediately by low-wage workers affected by tese policies. At least some of te burden of te increased wage bills faced by low-wage firms is passed on to te consumer troug iger prices. 2.3 Reducing Profits Since te minimum wage forces employers to pay iger wages, many policymakers and voters presume tat minimum wages will be paid out of employer profits. However, a variety of reasons lead one to suspect tat profits will not be a significant source for paying te costs of minimum wages. Most economic teory does not suggest tat profits are a likely source of covering costs. Rebitzer and Taylor (1995), for example, sow in a simple employment matcing model wit a large number of employers tat te introduction of a minimum wage does not reduce profits for employers. Also, Card and Krueger (1995) demonstrate tat te introduction of a minimum wage in an efficiency wage model does not reduce profits for employers. From a less formal perspective, low-wage employers are less likely tan oter employers to ave large profits. Te firms tat typically employ low-wage workers are in igly competitive industries. Income tax return data for major industries tat employ low-wage workers (e.g., food stores, eating and drinking establisments, retail trade and department stores) sow tat most of tese industries ave lower net incomes tan te average across all industries. 5 Low-wage workers are also more likely to work for small employers (e.g., see Card and Krueger (1995)). Small employers face greater competition in bot te labor market and te product market, meaning tat tey are unable to command monopoly power in te iring of workers or in te setting of product prices, and terefore ave lower profits. Moreover, even among te most profitable firms, capital is unlikely to bear te costs of a wage increase. Tis is especially true for large, publicly traded firms. It is a general result in public finance tat taxes are borne by tose wo are least able to adjust. Capital stock markets are extremely efficient, and te supply of capital is very price sensitive, meaning tat a small decrease in returns to capital will cause investors to move teir money into a firm wit better returns. Firms terefore cannot reduce te returns on teir stock and still expect investment. 5 Source: Internal Revenue Service, Corporate Income Tax Returns,

11 Unfortunately, little empirical researc exists on tis subject. Card and Krueger (1995) use an event study of stock prices of firms tat employ many low-wage workers suc as McDonald's and Wal-Mart. However, stock prices follow investors expectations about future profitability, so te connection between stock prices and te minimum wage is tenuous at best. Card and Krueger find little systematic relationsip between excess returns and news about minimum wage canges. Using data from te United Kingdom, Draca, Macin, and Van Reenen (2011) find some evidence suggesting tat te minimum wage reduces firm profits in te very sort run, but te long run impacts are left unanswered. Tus, despite te popular belief tat firms pay for minimum wage increases troug lower profits, tere is little empirical evidence to date supporting tis ypotesis, and basic economics suggests compelling reasons wy tis would not occur. In fact, te discussion of te GE model described later in tis paper outlines wy economic teory could predict tat returns to capital (and, tus, profits) can be expected to rise in response to an increase in a minimum wage wen employment losses are assumed not to occur for te labor receiving tis wage. 2.4 Assumptions on Paying for Minimum Wages in Assessing Distributional Consequences To depict te circumstances deemed most likely to apply by minimum wage advocates, te analysis below assumes tat no employment or profit losses occur as a result of minimum wage increases. Altoug many economists remain convinced tat increases in te minimum wage will decrease employment, te recent literature on tis subject as convinced most policymakers tat suc employment effects are very minimal. Wile many in te public policy community intimate tat minimum wage increases are paid out of firm profits, no reliable evidence supports tis position and few minimum wage advocates in te U.S. cite tis position. 6 Tis leaves price adjustments as te source for paying for minimum wage increases. If all te costs of te minimum wage are passed on to consumers in te form of iger prices, ten price increases sould reflect te wage increase multiplied by labor's sare of te total cost. In order to ave no job or profit loss, consumers must continue to purcase te same amount of low-wage goods at te iger price. Tus, our simulations make tree related assumptions: consumers do not reduce consumption as prices rise; all increased labor costs are passed on in iger prices; and low-wage workers remain employed at te same number of ours after te minimum wage rises. Taken togeter, tese tree assumptions provide a setting for simulating te expected effects of 6 If minimum wages do reduce profits, ten teir effects on te income distribution may be more progressive tan measured in tis study, since stock olders tend to be more wealty Americans. However, ow muc more progressive is unclear since many Americans, even ones tat are not particularly wealty, own stock troug private and public retirement portfolios. 10

12 minimum-wage increases in a relatively straigtforward manner. One need not believe tat all tese assumptions old in reality, preferring instead to believe tat firms pay for minimum wage ikes troug all possible sources. Tis simulation environment, owever, depicts a world wit no-job-loss wic is te notion popularly maintained by proponents of te minimum wage. Te simulation findings provide a basis for understanding te effectiveness of te minimum wage in redistributing resources across te ouseold income distribution. 11

13 3. Overview of Metodology and Data Altoug te above discussion primarily focuses on payment sources for costs, one must also consider te benefit side of te picture to understand te distributional effects of a minimum wage. Te two sides of te simulation analysis benefits and costs presented below require two different data sets. Tis section provides an overview tese data and te metodology applied to measure te benefits and costs of an increase in te minimum wage. 3.1 Description of Data To calculate te benefits of a minimum wage increase, te analysis relies on data from SIPP, a nationally representative survey of ouseolds conducted by te U.S. Census Bureau. To depict circumstances relevant to te 1996 increase in te federal minimum wage, tis analysis uses data from waves 1-3 of te 1996 SIPP, as te dates covered by tese survey waves place tem before te 1996 cange in minimum wage. Te SIPP data provides information on ouseolds, families, and individuals over 15 years of age, including montly data on income and earnings by source, wages, and ours worked, demograpic caracteristics, family structure, and public assistance program participation. Tese data permit identification of low-wage workers, teir occupations and industries, teir family income, and sufficient information to determine income tax burdens under alternative income scenarios using te National Bureau of Economics Researc (NBER) income tax simulator (TAXSIM) program. Te following analysis uses SIPP to simulate bot te before- and after-tax effects of a minimum wage increase on te incomes of families wit various caracteristics. To relate price increases to a family s purcases, te analysis relies on data from te CES matced to te same time period as te SIPP. Tis survey includes information on family expenditures on a wide variety of goods and services. It also incorporates a number of income measures and demograpic caracteristics, including family structure. Altoug te income and demograpic measures are not as precise as tose in SIPP, bot data sets allow identification of te same major categories of families, suc as teir position in te income distribution, poverty level, and welfare status. To trace te iger earnings of workers affected by te minimum wage to te products tey produce and ten to te consumers of tese products, te analysis uses national input-output tables. Tese tables are constructed by te Minnesota IMPLAN Group from databases on employment, value added, output, and product demand for 528 industrial sectors in all states and counties in te U.S. 7 7 Te IMPLAN data comes from data collected by te U.S. Bureau of Economic Analysis, te U.S. Bureau of Labor Statistics, and te U.S. Census Bureau, among oter sources. 12

14 3.2 Overview of Metodology Figure 1 illustrates te steps comprising te metodology implemented below to simulate te distributional consequences of increases in te minimum wage. In te figure, datasets are listed in bold font and te arrows indicate inputs into te next step. Using SIPP data, te first step calculates te effect of te 1996 increase in te minimum wage on te earnings of affected workers and teir family income. Tis is done assuming no cange in ours worked. Section 4.1 describes te precise formulation for tese calculations. Tis information is used bot for te benefit and te cost sides of te computations. On te benefits side, tese SIPP calculations measure ow muc eac individual family in te survey benefit from te wage increase. Te second step computes te distribution of tese benefits across families categorized by teir income quintiles, poverty levels, extent of dependence on low-wage earnings, welfare recipient status, and demograpic caracteristics. To translate benefits into after-tax values, te tird step applies te NBER TASXIM calculator to eac family s circumstances to determine ow muc of tese additional benefits (i.e., earnings) are reduced troug federal, state, and payroll taxes. Tis produces te final after-tax benefits for eac family. Te last step generates te distribution of after-tax benefits for te same family categorizations as used for te before-tax distributions. Section 4 presents tese findings. Computations for te cost side of te minimum wage increase are far more callenging. Inferring te sares of costs borne by te different categories of families requires two sets of calculations: (i) measures ow muc prices rise by commodity in response to te minimum wage increase, and (ii) te effects of tese price increases on consumption costs by family given its consumption composition across commodities. Computing price impacts require two steps after te first step described above comprised of te SIPP calculations measuring ow muc te labor cost of eac individual family in te survey rises due to a minimum wage increase. Using information in SIPP on eac low-wage worker s industry of employment, te second step computes te amounts tat labor costs rise in eac industry. In addition to iger wage costs, employers must also pay iger payroll taxes, primarily in te form of employers contributions to Social Security. Bot iger wages and taxes are included in te increased labor costs computed by industry. Ten, using te IMPLAN input-output tables, te tird step translates tese iger employment costs (i.e., direct costs) into price increases for eac final consumer good and service. Tis is simply an accounting exercise consistent wit te assumption tat firms respond to iger labor costs by increasing prices. Section 5.1 presents details on using te input-output tables to calculate final price increases. Turning to te consumption costs, and building off steps one troug tree above, te fourt step uses data from te CES to identify te composition and levels of consumption by different family types for 13

15 eac good and service. Te fift step translates te price increases calculated in previous steps into cost increases for eac consumer product. Families in te CES are categorized by income quintiles, consumption quintile, poverty status, welfare participation, and oter family caracteristics. Te sixt and last step computes te additional cost to eac family, assuming no cange in te family s consumption beavior, by combining te information on te distribution of consumption wit te implied price increases for eac commodity bundle. Section 5.3 presents tese findings. Finally, to infer te net effects of an increase in minimum wages, Section 6 integrates te benefits and cost allocations across and witin family types to compute te overall distributions for eac category of families. Te analysis also calculates te aggregate benefits and costs transfers troug a minimum wage increase. 14

16 4. Wo Benefits from Increases in te Minimum Wage? Tis section first sows ow to calculate te additional pre-tax and post-tax earnings for eac family induced by an increase in te minimum wage, and ten examines ow tese additional earnings are distributed across families by a variety of caracteristics wit empasis on particular types of families tat migt be considered te most important targets of minimum wage policy. Lastly, te section reviews previous researc done on te distribution of benefits. 4.1 Calculating Pre- and After-Tax Benefits of Families wit Low-Wage Workers Family gross earnings and income are raised by te combined increase in earnings of all family members; tis cange in family earnings is te pre-tax benefit and is calculated as follows. For eac worker in te family identified as earning an ourly wage below te new legally specified minimum wage level in 1996, te analysis assumes is or er ourly wage rises to te new minimum, tat is, from as low as $4.25 (te old minimum) to exactly $5.15 (in 1996 dollars). Te computations use te new wage rate and annual number of ours worked to calculate te implied increase in total earnings for eac worker during te year assuming tat tere is no cange in ours of work. For workers earning less tan te old minimum wage of $4.25, te analysis assumes tat tey also receive a $0.90 wage increase, wic does not bring tem up to te full $5.15 per our. Te computations assume no spillover benefits for workers already earning more tan te new minimum wage. For te after-tax benefit, te analysis adjusts te increased income for federal and state income taxes (including te Earned Income Tax Credit (EITC)) as well as payroll taxes using te NBER TAXSIM program. Tese calculations account for te dependent status of young workers as tis plays an important role in determining tax liability. 8 Tese calculations also assume tat all married couples are joint taxpayers. Because of data limitations, all taxpayers are assumed to take te standard deduction rater tan itemize teir deductions, wic sould ave little impact on low-income taxpayers. 4.2 Distribution of Benefits across Families by Income: Before and After Tax Using te before- and after-tax benefits calculated for eac family in SIPP, one can compute te sares of benefits received by families sorted by a variety of caracteristics, including income quintiles, income as a multiple of te poverty level, presence of cildren, eadsip and marriage status, wage rate 8 In 1996, taxpayers could claim a dependent exemption if tey ad a dependent under age 18 or ad a dependent under age 23 wo was a full time student. Te computations ere assume tat any cild under age 18 tat lived at ome for some part of te sample period and earned less tan $20,000 (in 1996 dollars) was claimed as a dependent by te parent(s). Cildren under age 23 wo reported being enrolled in college were also assumed to be claimed as dependents by te parent(s). Te TAXSIM program fully accounts for tese factors in its calculations of income taxes and EITC. 15

17 levels, and dependency on public assistance. Table 1 presents te distributions of benefits across different partitions of families. To igligt te distribution of benefits across family income, te top set of rows of Table 1 segments families into five income quintiles and reports te average levels and distribution of benefits (i.e., iger earnings) across tese quintiles. For eac quintile, column (5) sows te sare of families tat includes one or more minimum wage workers (i.e., tose wo benefit from te minimum wage increase). Te result is peraps surprising for tose unfamiliar wit similar findings in te literature. Te minimum wage population is almost perfectly distributed across te income distribution. Wile 22.3% of all families ave one or more minimum wage workers, only sligtly more (22.6%) of families in te lowest quintile include low-wage workers and terefore benefit from te minimum wage increase. Tis is nearly identical to te 22.7% of families in te igest income quintile tat ave a worker wo benefits from a minimum wage increase. Tus approximately one in five families benefit, regardless of teir income. Te more relevant question of were do te dollars go? is addressed in columns (2) troug (4) of Table 1. If ig-income ouseolds ave low-wage workers wo typically work fewer ours tan te lowwage workers at te bottom of te distribution (e.g., part-time teenagers as opposed to family breadwinners), ten one would expect te additional dollars from te wage increase to flow disproportionately to te poorer families. Column (2) presents te distribution of additional earnings due to te minimum wage increase across te five quintiles. If te benefits were identical distributed across all families, eac quintile would receive about 20% of te extra earnings, and more tan its sare of te additional earnings if it receives more tan 20%. Tis is essentially te story revealed in Table 1: benefits are evenly divided across quintiles. Te 40% of families at te bottom of te income distribution receive only 38.3% of te additional earnings from te minimum wage. Conversely, te top 40% of families receive 40.3% of te extra earnings. Te minimum wage increase distributes money to families at all income levels wit little preference given to any group. Since te U.S. tax system is progressive, te distribution of extra earnings canges wen calculating te sares of earnings after taxes, as reported in column (3). Te poorest families lose less of teir extra earnings to taxes: teir sare drops only 2.2 points from 19.9% to 17.7%. Tose families in te igest income quintile fare worse: teir sare drops 6 percentage points from 18.6% to 12.6%. Te distributional impact of te tax system is also apparent from comparing te average value of after-tax benefits for families tat ave a minimum wage worker as reported in Column (4) of Table 1. Again, low-income families benefit more tan ig-income families, toug not by as muc as migt ave been expected. Troug taxation, te government captures about one quarter of te total benefits from te minimum wage increase. 16

18 Tese calculations ignore te potential loss of cas and in-kind welfare benefits for families under and near te poverty level wose income rises due to te minimum wage. Te computation of after-tax benefits performed in tis analysis includes transfers from te EITC program, but not from suc income support programs as Temporary Assistance to Needy Families (TANF), Aid to Families wit Dependent Cildren (AFDC), and food stamps. Accounting for tese welfare transfers would strictly worsen te distributional consequences of te minimum wage conveyed by tis study. 4.3 Benefits to Oter Target Families Wile ranking families by income does not take into account family size, poverty levels do. Te tird set of rows in Table 1 report te sares of minimum wage benefits going to families wit income and sizes measured against multiples of te poverty tresold. As sown in te after-tax sares in Table 1, 13.4% of benefits go to families below te poverty tresold. However, nearly 30% of te after-tax benefits go to families wit incomes tat are more tan tree times te poverty tresold. Tus, te majority of te additional earnings do not go to poor (or near poor) families. Anoter primary target of te minimum wage consists of families dependent on te earnings from a low-wage worker for a substantial part of total family earnings. Te fourt set of rows in Table 1 lists results for four different specifications of families wit cildren tat rely on te earnings of low-wage employees: families for wic more tan 50% of teir total earnings come from employment paying (i) no more tan $5.15 per our, (ii) no more tan $6.00 per our, (iii) no more tan $7.50 per our, and (iv) no more tan $10.00 per our. Not surprisingly, Table 1 sows tat tese target families receive larger after-tax benefits on average and receive a disproportionate sare of minimum wage benefits. For example, families in te tird category above receive 20% of all minimum wage benefits, even toug tey make up only 7% of all families. However, even wen te low-wage tresold is expanded to include wages as ig as $10.00 per our, only 22% of total after-tax minimum wage benefits go to tese target families. Te last set of rows in Table 1 present projected allocations for married and single families, distinguising tose wit cildren. In general, families wit cildren receive more benefits tan tose witout. Table 1 also gives results for families wo received welfare at some time during te year. Interpreting welfare as public cas aid and/or food stamps, welfare recipient families wit cildren account for 9.5% of families and tey are projected to receive 13.8% of te after-tax additional earnings generated by a minimum wage increase. 4.4 Previous Researc on te Distribution of Benefits Tis assessment of te distribution of benefits mostly replicates early work by Gramlic (1976), 17

19 Jonson and Browning (1983), Burkauser and Finegan (1989), Horrigan and Mincy (1993), and Burkauser and Sabia (2007). Tese studies also document tat many low-wage workers are members of ig-income families. Tis is especially true for teenagers wo are distributed trougout te entire family income distribution and often find employment in minimum wage jobs. Tis literature consistently sows tat wile te minimum wage as a small effect on earnings inequality, it as virtually no effect on income inequality. 9 Jonson and Browning (1983) and Horrigan and Mincy (1993) focus on te distribution of minimum wage benefits by family income quintile and sow tat te additional minimum wage earnings are only mildly redistributive, wit somewat larger benefits going to families in te second to lowest income quintile. Burkauser and Finegan (1989) and Burkauser, Couc and Wittenburg (1996) focus on te distribution of benefits by families income measured as multiples of te poverty tresold. Tey find tat te distribution of benefits is not significantly different from te population sares. Burkauser and Finegan (1989), for example, find tat only 18% of workers wo benefit from a minimum wage increase ad a family income tat was below te poverty tresold. Burkauser, Couc and Wittenburg (1996) find tat only 13% of affected workers were in poverty. Card and Krueger (1995) report similar results, as do Burkauser and Sabia (2007) wic reports benefits sares not only on te distribution of minimum wage benefits by family income quintile, but also for near-poor families defined by poverty levels. 4.5 Summary: Distribution of Benefits Minimum wage policy offers an inefficient mecanism for boosting te incomes of families tat policymakers typically tink of as te intended beneficiaries of minimum wage increases: poor families, tose supported primarily by low-wage work, and tose on welfare. About 35% of te total increase in after-tax benefits goes to families wit income less tan two times te poverty tresold, a common definition of te working poor or near poor; nearly 13% goes to families principally supported by low-wage workers defined as earning wages at or below 117% (=$6.00/$5.15) of te new 1996 minimum wage; and only about 14% goes to families wit cildren on welfare. Unlike most public income support programs, increased earnings from te minimum wage are taxable. Over 25% of te increased earnings are collected back as income and payroll taxes, including te net effect of EITC wic subsidizes low-earning families. Even after taxes, 27.6% of increased earnings go to families in te top 40% of te income distribution. 9 Several sets of results in Table 1 are not elsewere in te literature: most important, benefits going to families wic depend on low-wage employment for more tan alf of total family earnings and to families wo participate in a welfare program. Te findings for tese groups, owever, fit wit te well-establised conclusion of tis literature: te minimum wage represents a very blunt policy instrument for providing benefits to low-income families. 18

20 5. Wo Pays for Increases in te Minimum Wage? If employment and profits are unaffected, ten te cost of te minimum wage increase is covered troug iger prices. As prices rise on te goods and services produced by low-wage workers, all consumers of tese products are essentially subsidizing te low-wage workers. Te following discussion sows tat prices rise on a wide variety of goods, imposing across-te-board price increases tat it all consumers. To assess te distributional impacts of tese price increases, Section 5.1 relies on national inputoutput tables to calculate ow muc individual product prices must rise to cover te new labor costs induced by te minimum wage increase, and Section 5.2 summarizes te findings produced by tis analysis. From te employer's perspective, te increase in labor costs will be greater tan te increase in earnings since employers will also ave to pay iger payroll tax contributions. Tese price calculations assume a national market wit te new prices imposed on all consumers. Te analysis ten translates tese price increases into total consumption cost by family, and Section 5.3 describes te allocation of tese consumption costs across families broken down by teir income and demograpic caracteristics. 5.1 Attributing Labor Costs to Price Increases Te first step in determining wo pays for te minimum wage ike involves calculating te impact of te increased labor costs on te total cost of final goods and services. Te following analysis assumes tat, if te cost of labor increases in a particular industry, ten te price of tat industry s output will rise to increase consumer expenditures by te same amount. Tere are two ways for te total cost of goods to increase after a minimum wage increase. First, tere is te direct effect on te cost of labor for industries iring low-wage workers. Second, tere is te indirect effect troug intermediate goods. Wile some portion of an industry's output is consumed by final users (e.g. ouseolds and government), te rest of te output is allocated to intermediate use, were te output of te original industry becomes an input for anoter. Tus, even if an industry employs no minimum wage workers, te prices for tat industry's output may rise because te industry uses goods or contracts for services produced wit minimum wage labor. Tis feedback troug intermediate uses continues ad infinitum, so te price sock from te wage ike propagates trougout te economy. Te calculations begin by determining te industries tat employ low-wage workers. From te SIPP, one can identify all industries tat employed workers at wages below te new minimum of $5.15. Considering all low-wage workers in a given industry, one can infer te total increase in industry labor costs, including additional employer contributions for Social Security, resulting from te wage ike. Denote 19