Minimum Wages and Employment_ A Case Study of the Fast-Food Industry in New Jersey and Pennsylvan

Minimum Wages and Employment: A Case Study of the Fast-Food Industry in New Jerseyand Pennsylvania

Author(s): David Card and Alan B. Krueger

Source: The American Economic Review, Vol. 84, No. 4 (Sep., 1994), pp. 772-793Published by: American Economic AssociationStable URL: http://www.jstor.org/stable/2118030Accessed: 08/09/2010 23:12

Your use of the JSTOR archive indicates your acceptance of JSTOR's Terms and Conditions of Use, available at

http://www.jstor.org/page/info/about/policies/terms.jsp. JSTOR's Terms and Conditions of Use provides, in part, that unlessyou have obtained prior permission, you may not download an entire issue of a journal or multiple copies of articles, and youmay use content in the JSTOR archive only for your personal, non-commercial use.

Please contact the publisher regarding any further use of this work. Publisher contact information may be obtained athttp://www.jstor.org/action/showPublisher?publisherCode=aea.

Each copy of any part of a JSTOR transmission must contain the same copyright notice that appears on the screen or printedpage of such transmission.

JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range ofcontent in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new formsof scholarship. For more information about JSTOR, please contact support@jstor.org.

American Economic Association is collaborating with JSTOR to digitize, preserve and extend access to TheAmerican Economic Review.

http://www.jstor.org

Minimum Wages and Employment: A Case Study of the Fast-Food Industry in New Jersey and Pennsylvania By DAVID CARD AND ALAN B. KRUEGER* On April 1, 1992, New Jersey's minimum wage rose from $4.25 to $5.05 per in 410 fast-food restaurants hour. To evaluate the impact of the law we surveyed New Jersey and eastern Pennsylvania before and after the rise. Comparisons of employment growth at stores in New Jersey and Pennsylvania (where the minimum wage was constant) provide simple estimates of the effect of the higher minimum wage. We also compare employment changes at stores in New Jersey that were initially paying high wages (above $5) to the changes at lower-wage stores. We find no indication that the rise in the minimum wage reduced employment. (JEL J30, J23) How do employers in a low-wage labor market respond to an increase in the mini- mum wage? The prediction from conven- tional economic theory is unambiguous: a rise in the minimum wage leads perfectly competitive employers to cut employment (George J. Stigler, 1946). Although studies in the 1970's based on aggregate teenage employment rates usually confirmed this prediction,1 earlier studies based on com- parisons of employment at affected and un- affected establishments often did not (e.g., Richard A. Lester, 1960, 1964). Several re- Department of Economics, Princeton University, Princeton, NJ 08544. We are grateful to the Institute for Research on Poverty, University of Wisconsin, for partial financial support. Thanks to Orley Ashenfelter, Charles Brown, Richard Lester, Gary Solon, two anonymous referees, and seminar participants at Princeton, Michigan State, Texas A&M, University of Michigan, University of Pennsylvania, University of Chicago, and the NBER for comments and sugges- tions. We also acknowledge the expert research assis- tance of Susan Belden, Chris Burris, Geraldine Harris, and Jonathan Orszag. 1See Charles Brown et al. (1982, 1983) for surveys of this literature. A recent update (Allison J. Wellington, 1991) concludes that the employment effects of the minimum wage are negative but small: a 10-percent increase in the minimum is estimated to lower teenage emplovment rates by 0.06 percentage noints. 772 * compara- cent studies that rely on a similar tive methodology have failed to detect a effect of higher mini- negative employment mum wages. Analyses of the 1990-1991 in- creases in the federal minimum wage 1992; Card, and Krueger, (Lawrence F. Katz 1992a) and of an earlier increase in the minimum (Card, 1992b) wage in California A study employment impact. find no adverse in Britain floors (Stephen of minimum-wage reaches a 1994) Machin and Alan Manning, similar conclusion. This paper presents new evidence on the effect of minimum wages on establishment- the We analyze level employment outcomes. in of 410 fast-food restaurants experiences the New Jersey and Pennsylvania following increase in New Jersey's minimum wage from $4.25 to $5.05 per hour. Comparisons of employment, wages, and prices at stores in New Jersey and Pennsylvania before and after the rise offer a simple method for evaluating the effects of the minimum wage. within New Jersey between Comparisons (those paying more initially high-wage stores than the new minimum rate prior to its effective date) and other stores provide an alternative estimate of the impact of the new law. In addition to the simplicity of our empir- ical methodology, several other features of VOL. 84 NO. 4 AND EMPLOYMENT CARD AND KRUEGER: MINIMUM WAGE 773 the New Jersey law and our data set are also significant. First, the rise in the mini- a recession. The mum during wage occurred increase had been legislated two years ear- lier when the state economy was relatively By the time of the actual increase, healthy. the unemployment rate in New Jersey had political risen substantially and last-minute action almost succeeded in reducing the of the relative mini- and 1991 to measures mum wage in each state. I. The New Jersey Law A bill signed into law in November 1989 $3.35 minimum wage from raised the federal per hour to $3.80 effective April 1, 1990, to $4.25 per hour on increase with a further minimum-wage increase. It is unlikely the effects of the higher minimum wage that were obscured by a rising tide of general economic conditions. state with an economy Second, New Jersey is a relatively sto nearby group sthat is closely linked mall vania forms a natural of fast-food tates. We believe that a control stores in eastern Pennsyl- with the experiences Jersey. obf restaurants in New asis for comparison across stores Jersey, however, Wage variation allows us to compare in New the experiences stores within New Jersey of high-wage and low-wage and to validity control test the Moreover, since of the Pennsylvania seasonal patterns of em- group. ployment are similar in New Jersey and eastern Pennsylvania, high- and low-wage tively \"differences our comparative methodology stores within New Jer- as well as across sey, effec- out\" any seasonal em- ployment effects. percent Third, we successfully followed nearly 100 views conducted of stores from a first minimum wage (in February just before the rise wave of inter- and March in the 1992) to a second wave conducted months after (in November and December 7-8 1992). store closings and take account We have complete information on ment changes at the closed stores of employ- analyses. effect of the minimum We therefore mthe overall in our employment, and not simply weasure age on average its effect on surviving establishments. - Our analysis of employment trends at stores that were open for business the increase in the minimum any potential wage ignores before the rate of new store openings. To assess effect of minimum wages on the likely state-specific growth magnitude of this effect we relate McDonald's fast-food rates in the number outlets between 1986 of April 1, 1991. legislature parallel went one step further, In early 1990 the New Jersey enacting for 1990 increases in the state per hour effective and 1991 and an increase minimum wage uled 1992 increase gave New Jersey the April 1, 1992. The sched- to $5.05 highest and was strongly state minimum wage in the country opposed by business of National lead- ers in the state (see Bureau Affairs, Daily Labor Report, In the two years between passage 5 May 1990). of the $5.05 minimum wage New Jersey's economy slipped and its effective date, sion. Concerned with the potentially ad- into reces- verse state legislature voted in March 1992 to impact of a higher minimum wage, the phase The vote fell just short of the margin in the 80-cent increase over two years. quired the Governor to override a gubernatorial veto, and re- into effect on April 1 before vetoing the allowed the $5.05 rate to go two-step of having legislation. Faced wage earners, the legislature to roll back wages for minimum- with the prospect issue. Despite a strong last-minute dropped the lenge, the $5.05 minimum as originally planned. rate took effect chal- II. Sample Design and Evaluation impending increase Early in 1992 we decided to evaluate mum in the New Jersey mtini- he rants nia.2 Our choice of the fast-food industry iwage by surveying n New Jersey and eastern fast-food restau- Pennsylva- was driven by several stores are a leading employer factors. First, fast-food workers: in 1987, franchised restaurants em- of low-wage 2At the time we were uncertain whether the $5.05 rate would go into effect or be overridden. 774 THE AMERICAN ECONOMIC REVIEW TABLE 1-SAMPLE DESIGN AND RESPONSE RATES SEPTEMBER 1994 Stores in: All Wave 1, February 15-March 4, 1992: Number of stores in sample frame:a Number of refusals: Number interviewed: Response rate (percentage): Wave 2, November 5-December 31, 1992: Number of stores in sample frame: Number closed: Number under rennovation: Number temporarily closed:b Number of refusals: Number interviewed:c 410 6 2 2 1 399 331 5 2 2 1 321 79 1 0 0 0 78 473 63 410 86.7 364 33 331 90.9 109 30 79 72.5 NJ PA aStores with working phone numbers only; 29 stores in original sample frame had disconnected phone numbers. bIncludes one store closed because of highway construction and one store closed because of a fire. CIncludes 371 phone interviews and 28 personal interviews of stores that refused an initial request for a phone interview. ployed 25 percent of all workers in the restaurant industry (see U.S. Department of table 13). fast-food Commerce, 1990 Second, restaurants comply with minimum-wage reg- ulations and would be expected to raise wages in response to a rise in the minimum wage. Third, the job requirements and products of fast-food restaurants are rela- tively homogeneous, making it easier to ob- tain reliable measures of employment, wages, and product prices. The absence of of tips greatly simplifies the measurement it is relatively wages in the industry. Fourth, easy to construct a sample frame of fran- chised restaurants. Finally, past experience (Katz and Krueger, 1992) suggested that fast-food restaurants have high response rates to telephone surveys.3 Based on these considerations we con- structed a sample frame of fast-food restau- rants in New Jersey and eastern Pennsylva- nia from the Burger King, KFC, Wendy's, and Roy Rogers chains.4 The first wave of the survey was conducted by telephone in late February and early March 1992, a little over a month before the scheduled increase in New Jersey's minimum wage. The survey included questions on employment, starting wages, prices, and other store characteris- tics.5 Table 1 shows that 473 stores in our sam- ple frame had working telephone numbers when we tried to reach them in February- March 1992. Restaurants were called as many as nine times to elicit a response. We obtained completed interviews (with some item nonresponse) from 410 of the restau- rants, for an overall response rate of 87 percent. The response rate was higher in New Jersey (91 percent) than in Pennsylva- very low response rates from McDonald's restaurants. For this reason, McDonald's restaurants were excluded from Katz and Krueger's and our sample frames. 3In a pilot survey Katz and Krueger (1992) obtained 4The sample was derived from white-pages tele- phone listings for New Jersey and Pennsylvania as of February 1992. 5Copies of the questionnaires used in both waves of the survey are available from the authors upon request. VOL. 84 NO. 4 CARD AND KRUEGER: MINIMUM WAGE AND EMPLOYMENT 775 nia (72.5 percent) because our interviewer made fewer call-backs to nonrespondents in Pennsylvania.6 In the analysis below we in- vestigate possible biases associated with the degree of difficulty in obtaining the first- wave interview. The second wave of the survey was con- ducted in November and December 1992, about eight months after the minimum-wage nently closed stores but is treated as missing for the temporarily closed stores. (Full- time-equivalent [FTE] employment was cal- culated as the number of full-time workers [including managers] plus 0.5 times the number of part-time workers.)8 Means are presented separately for stores in New Jer- sey and Pennsylvania, along with t statistics for the null hypothesis that the means are increase. Only the 410 stores that re- sponded the second round in the first wave were contacted in fully interviewed ostores of a concern by phone in November 3f interviews. We success- 71 (90 percent) of these might have closed, we hired an interviewer that nonresponding restaurants 1992. Because to drive to each of the 39 nonrespondents and determine open, and to conduct whether the store was still possible. a personal interview if restaurants The interviewer discovered that were temporarily closed (one because of a were permanently closed, two six fire, one because of road two open for business, all but one granted a were under renovation.7 construction), and Of the 29 stores request for a personal interview. sult, we for 99.8 percent of the restaurants that re- have second-wave interview data As a re- sponded information in the first wave of the survey, and cent of the sample. on closure status for 100 per- key variables Table 2 presents the means for several the subset in our data set, averaged variable. ment in wave 2 is set to 0 for the perma- In constructing the of nonmissing responses means, feor each over mploy- in the two states. Among New Jersey stores, 44.5 6Response rates per call-back were almost identical percent responded responded sylvania stores after at most two call-backs. Among on the first call, and 72.0 percent and 71.6 percent responded 42.2 percent responded on the first Penn- backs. after at most two call- call, construction and one of the stores closed for renova- 7As of April 1993 the store closed because of road tion had reopened. when our telephone interviewer called in November The store closed by fire was open 1992 but refused the interview. follow-up personal store. interview a mall By the time of the fire had closed the equal in the two states. by chain and ownership status (company- Rows la-e show the distribution of stores owned versus franchisee-owned). The Burger King, Roy Rogers, and Wendy's stores in our sample have similar average food prices, store hours, and employment levels. The KFC stores are smaller open for fewer hours. They also offer a and are more expensive the other chains m(chicken ain course than stores in vs. hamburgers). full-time equivalent workers per store in In wave 1, average employment was 23.3 Pennsylvania, compared 20.4 in New Jersey. Starting wages were with an average of very similar although among s(medium soda, the average tores in the two states, was were no significant cross-state significantly small phigher fries, rice of a \"full meal\" in New and an Jersey. entree) There average full-time workers, or the prevalence of bonus hours of operation, the fraction differences in of programs to recruit restaurants The average starting wage at fast-food new workers.9 percent following in New Jersey increased by 10 wage. Further insight into this change is the rise in the minimum provided tributions of starting wages in Figure 1, which shows the dis- before and after the rise. In wave 1, the in the two states distributions in New Jersey and Pennsylva- nia were very similar. By wave 2 virtually all tive assumptions on the measurement of employment 8We discuss the sensitivity of our results to alterna- in Section I\"bounty\" 9These programs II-C. on the job for a minimum for recruiting any new employee offer current employees a cash wbounties the recruiter with an \"employee are $50-$75. Recruiting programs that period of time. Typical ho stays anation tabulations. or other noncash bonuses aore f the month\" excluded from dward esig- our 776 THE AMERICAN ECONOMIC REVIEW SEPTEMBER 1994 TABLE 2-MEANS OF KEY VARIABLES Stores in: Variable NJ PA ta 1. Distribution of Store Types (percentages): a. Burger King 41.1 44.3 -0.5 b. KFC 20.5 15.2 1.2 c. Roy Rogers 24.8 21.5 0.6 d. Wendy's 13.6 19.0 - 1.1 e. Company-owned 34.1 35.4 -0.2 2. Means in Wave 1: a. FTE employment 20.4 23.3 -2.0 (0.51) (1.35) b. Percentage full-time employees 32.8 35.0 -0.7 (1.3) (2.7) c. Starting wage 4.61 4.63 -0.4 (0.02) (0.04) d. Wage = $4.25 (percentage) 30.5 32.9 -0.4 (2.5) (5.3) e. Price of full meal 3.35 3.04 4.0 (0.04) (0.07) f. Hours open (weekday) 14.4 14.5 -0.3 (0.2) (0.3) g. Recruiting bonus 23.6 29.1 - 1.0 (2.3) (5.1) 3. Means in Wave 2: a. FTE employment 21.0 21.2 -0.2 (0.52) (0.94) b. Percentage full-time employees 35.9 30.4 1.8 (1.4) (2.8) c. Starting wage 5.08 4.62 10.8 (0.01) (0.04) d. Wage = $4.25 (percentage) 0.0 25.3 (4.9) e. Wage = $5.05 (percentage) 85.2 1.3 36.1 (2.0) (1.3) f. Price of full meal 3.41 3.03 5.0 (0.04) (0.07) g. Hours open (weekday) 14.4 14.7 -0.8 (0.2) (0.3) h. Recruiting bonus 20.3 23.4 -0.6 (2.3) (4.9) Notes: See text for definitions. Standard errors are given in parentheses. aTest of equality of means in New Jersey and Pennsylvania. restaurants paying starting less than $5.05 per hour reported in New Jersey that had been a equivalent employment increased in New Despite the increase in wages, full-time- estingly, wJersey relative to Pennsylvania. apparent the minimum-wage age equal to the new rate. Inter- \"spillover\" on higher-wage restau- increase had no New Jersey stores were initially smaller, Whereas rants employment change in the state: for these stores the mean was -3.1 percent. percentage wage with losses in Pennsylvania led to a small gains in New Jersey coupled and statistically insignificant interstate VOL. 84 NO. 4 CARD AND KRUEGER: MINIMUM WAGE AND EMPLOYMENT February 1992 35- 30- 25 - 0 0 15- 0. 10 5- 0 4.25 4.35 . 4.45 4.55 6L. 4.65 4.75 4.85 4.95 5.05 5.15 5.25 5.35 5.45 5.55 Wage Range November 1 9 P9y2 90 80- - 70 0 30- 0- 10 4.2 5 4.3 5 4.4 5 4.5 5 4.6 5 4.7 5 4.8 5 4.9 5 5.0 5 5.1 5 5.2 5 5.3 5 5.4 5 5.5 5 Wage Range New Jersey - Pennsylvania FIGURE 1. DISTRIBUTION OF STARTING WAGE RATES 777 778 THE AMERICAN ECONOMIC REVIEW SEPTEMBER 1994 difference ables show a relative in wave 2. Only two other vari- 1 and 2: the fraction change between waves and the price of a meal. Both variables of full-time employees increased in New Jersey relative to Pennsyl- vania. questionnaire We can assess the reliability of our survey of 11 stores that were inadvertently inter- by comparing the responses viewed Assuming that measurement twice in the first wave of the survey.10 two interviews are independent of each errors in the other and independent the correlation estimate of the \"reliability ratio\" between responses of the true variable, of the variance of the signal to the com- (the ratio gives an bined estimated reliability variance of the signal and noise). The ranging employment to 0.98 from 0.70 for full-time equivalent ratios are fairly high, for the price of a meal.1\" missing data for any key variables We have also checked whether stores with ferent from restaurants sponses. We find that stores with missing with complete re- are dif- data on employment, similar plete data. There is a significant size differ- in other respects wages, or prices are to stores with com- ential associated store closing after wave 1. The six stores with the likelihood of the that closed were smaller (with an average employment than other stores full-time-equivalent employees oin wave f only 12.4 1).12 III. Employment Effects of the Minimum-Wage Increase A. Differences in Differences changes Table 3 summarizes in average the levels and employment per store in 10These restaurants were interviewed twice because their phone numbers appeared in more than one phone book, and neither the interviewer nor the respondent noticed that they were previously interviewed. \"Similar reliability ratios for very similar questions were obtained by Katz and Krueger (1992). 12A probit analysis of the probability of closure shows that the initial size of the store is a significant predictor of closure. The level of starting wages has a numerically small and statistically insignificant coeffi- cient in the probit model. our survey. We present data by state in columns Jersey classified by whether the starting (i) and (ii), and for stores in New wage in wave 1 was exactly [column hour [column [column (iv)] between $4.26 and $4.99 per $4.25 per hour (v)] or $5.00 or more per hour in average (vi)]. and Pennsylvania employment between We also show the differences between stores in the various stores [column New Jersey (iii)] and in New Jersey wage ranges in average employment Row 3 of the table presents [columns (vii)-(viii)]. and 2. These entries are simply between waves 1 the changes ences between the averages for the two the differ- waves tive estimate of the change is presented in (i.e., row 2 minus row 1). An alterna- row 4: here we have computed in employment over that tthe change waves. We refer to this group of stores as reported valid employment data he subsample of stores in both the balanced sents the average subsample. Finally, row 5 pre- the balanced subsample, treating wave-2 change in employment in employment stores as zero, rather at the four temporarily closed than as missing. were initially As noted in Table 2, New Jersey stores nia counterparts but grew smaller than their Pennsylva- sylvania relative tmum wage. The relative gain (the \"dif- stores after the rise in the mini- o Penn- ference in differences\" employment) opercent), tion of the averages with a t statistic is 2.76 FTE employees f the changes in of 2.03. Inspec- (or 13 that the relative sey and tical when the analysis Pennsylvania stores change between New Jer- in rows 4 and 5 shows balanced subsample, iis virtually iden- smaller when wave-2 employment at the and it is only slightly s restricted to the temporarily closed stores is treated as panded Within New Jersey, employment ex- zero. $4.25 the high-wage stores (those paying per hour at the low-wage in wave 1) and contracted stores (those paying at more in employment at the high-wage stores per hour). Indeed, the average $5.00 or change (- 2.16 FTE employees) to the change among Pennsylvania is almost identical (- 2.28 FTE employees). Since high-wage stores stores in New Jersey should have been VOL. 84 NO. 4 CARD AND KRUEGER: MINIMUM WAGE AND EMPLOYMENT 779 largely unaffected by the new minimum wage, this comparison provides a specifica- tion test of the validity of the Pennsylvania els of the form: (la) AEi =a+bXj+cNJi+ e control group. The test is Regardless of whether the affected clearly passed. are compared high-wage stores in New Jersey, the to stores in Pennsylvania or stores mated employment effect of the minimum esti- wage is similar. ployment The results in Table 3 suggest that em- November contracted were unaffected of 1992 at fast-food stores between February and that wage (stores in Pennsylvania and by the rise in the minimum New Jersey in wave 1). We suspect paying $5.00 per hour or more stores in this contraction was the continued that the reason for ing of the economies states during 1992.13 oworsen- in New Jersey, Pennsylvania, and New York Uf the middle-Atlantic nemployment rates with a larger increase in all trended upward between 1991 Pennsylvania New Jersey than and 1993, franchised fast-food restaurants are pro- during 1992. Since sales of cyclical, expected to lower fast-food the rise in unemployment would be the absence of other factors.14 employment in B. Regression-Adjusted Models allowance The comparisons in Table 3 make employment growth, such as differences for other sources of variation no in across estimates in Table 4. The entries in chains. These are incorporated in the table are regression coefficients from mod- this '3An alternative possibility is that seasonal factors produce higher employment at fast-food restaurants in February and March than in November and December. An analysis of national employment data for food preparation and service workers, however, shows higher average employment in the fourth quarter than in the first quarter. 14To investigate the cyclicality of fast-food restau- rant sales we regressed the year-to-year change in U.S. sales of the McDonald's restaurant chain from 1976-1991 on the corresponding change in the unem- ployment rate. The regression results show that a 1-percentage-point increase in the unemployment rate reduces sales by $257 million, with a t statistic of 3.0. or (lb) AEj=a'+b'Xj+c'GAP1+ej where A from of characteristics of store i, and wave 1 to wave 2 at store i, Ei is the change in employment Xi is a set dummy New Jersey. variable that equals 1 for stores in NJi is a of the impact GAPi is an alternative measure i based on the initial wage at that store of the minimum wage at store GAPi = 0 for stores in Pennsylvania = 0 for stores in New Jersey with Wli 2 $5.05 = (5.05 - W1E)/ W1i for other stores in New Jersey. GAPi at store i necessary is the proportional increase in mum rate. Variation to meet the new wages in GAPE reflects mini- both the New Jersey-Pennsylvania differences within New Jersey contrast and based on re- ported starting wages in wave 1. Indeed, the value of of the actual proportional GAPi is a strong waves 1 and 2 wage change predictor between on (R2 = 0.75), and conditional behavior GAPj bthere is no difference in New in etween stores Jersey wage and Pennsylvania.15 The estimate in column (i) of Table 4 is directly comparable difference-in-differences of to the simple changes in column (iv), employment The row 4 of Table 3. discrepancy between the two estimates Table 4. In Table is due to the restricted 4 and the sample tin a- bles in this section remaining our to the set of stores we restrict ment and wage data with available in both waves of the eanalysis mploy- 15A regression of the proportional wage change be- tween waves 1 and 2 on GAPi has a coefficient of 1.03. 780 THE AMERICAN ECONOMIC REVIEW SEPTEMBER 1994 TABLE 3-AVERAGE EMPLOYMENT PER STORE BEFORE AND AFTER THE RISE IN NEW JERSEY MINIMUM WAGE Variable Stores by state Stores in New Jerseya Differences within NJb Difference, Wage = Wage = Wage 2 Low- Midrange- PA NJ - PA NJ $4.25 $4.26-$4.99 $5.00 high high (i) (ii) (iii) (vi) (iv) (v) (vii) (viii) -2.89 (1.44) -0.14 (1.07) 2.76 (1.36) 2.75 (1.34) 2.51 (1.35) 19.56 (0.77) 20.88 (1.01) 1.32 (0.95) 1.21 (0.82) 0.90 (0.87) 20.08 (0.84) 20.96 (0.76) 0.87 (0.84) 0.71 (0.69) 0.49 (0.69) 22.25 (1.14) 20.21 (1.03) -2.04 (1.14) -2.16 (1.01) - 2.39 (1.02) - 2.69 (1.37) 0.67 (1.44) 3.36 (1.48) 3.36 (1.30) 3.29 (1.34) -2.17 (1.41) 0.75 (1.27) 2.91 (1.41) 2.87 (1.22) 2.88 (1.23) 1. FTE employment before, 23.33 20.44 all available observations (1.35) (0.51) 2. FTE employment after, 21.17 21.03 all available observations (0.94) (0.52) 3. Change in mean FTE employment 4. Change in mean FTE balanced employment, sample of storesc 5. Change in mean FTE employment, setting FTE at temporarily closed stores to od -2.16 0.59 (1.25) (0.54) -2.28 0.47 (1.25) (0.48) - 2.28 0.23 (1.25) (0.49) Notes: Standard errors are shown in parentheses. The sample consists of all stores with available data on employment. FTE counts each part-time worker (full-time-equivalent) employment as half a full-time worker. at six closed stores Employment is set to zero. Employment at four temporarily closed stores is treated as missing. aStores in New Jersey were classified by whether starting wage in wave 1 equals $4.25 per hour (N = 101), is between $4.26 and $4.99 per hour (N = 140), or is $5.00 per hour or higher (N = 73). bDifference in employment between low-wage ($4.25 per hour) and high-wage ( 2 $5.00 per hour) stores; and difference in employment between midrange stores. ($4.26-$4.99 per hour) and high-wage CSubset of stores with available employment data in wave 1 and wave 2. dIn this row only, wave-2 employment at four temporarily closed stores is set to 0. Employment changes are based on the subset of stores with available data in wave 1 and wave 2. employment TABLE 4-REDUCED-FORM MODELS FOR CHANGE IN EMPLOYMENT Model Independent variable 1. New Jersey dummy 2. Initial wage gapa 3. Controls for chain and ownershipb (i) 2.33 (1.19) no no 8.79 (ii) 2.30 (1.20) - (iii) (iv) (v) yes no 8.78 0.34 15.65 (6.08) no no 8.76 - 14.92 (6.21) yes no 8.76 0.44 11.91 (7.39) yes yes 8.75 0.40 4. Controls for regionc 5. Standard error of regression 6. Probability value for controlsd Notes: Standard errors are given in parentheses. The sample consists of 357 stores with available data on employment and starting wages in waves 1 and 2. The dependent variable in all models is change in FTE employment. The mean and standard deviation of the dependent variable are -0.237 and 8.825, respectively. All models include an unrestricted constant (not reported). aProportional increase in starting wage necessary to raise starting wage to new minimum rate. For stores in Pennsylvania the wage gap is 0. bThree dummy variables for chain type and whether or not the store is company- owned are included. CDummy variables for two regions of New Jersey and two regions of eastern Pennsylvania are included. dProbability value of joint F test for exclusion of all control variables. VOL. 84 NO. 4 AND EMPLOYMENT CARD AND KRUEGER: MINIMUM WAGE 781 survey. smaller This restriction results in a slightly employment in New Jersey. estimate of the relative increase in set of four control variables: The model in column (ii) introduces a three of the chains dummies fcompany-owned probability values in row 6, these covariates stores. As shown by and another dummy the for or add little to the model and have no effect on the size of the estimated New Jersey dummy. the GAP variable The specifications in columns the minimum to measure the effect of (iii)-(v) use slightly sey dummy, although better fit than the simple New Jer- wage. This variable gives a New Jersey-Pennsylvania its implications for the similar. The mean value of comparison are New Jersey in column (iii) implies a 1.72 increase stores is 0.11. Thus GAPi the estimate among in FTE employment Pennsylvania. in New Jersey relative to New possible to add Since GAP varies Jersey, it is employment cient of the New Jersey dummy model. The estimated coeffi- both within GAPi and NJi to the vides a test of the Pennsylvania then pro- group. coefficient When we estimate these models, control the significant (with of the New Jersey dummy is in- ing that inferences about the t ratios of 0.3-0.7), imply- minimum wage are similar whether effect of the the comparison is made stores across states or across initial with higher and lower win New Jersey umn (v), where we have added dummies An even stronger ages. test is provided in col- representing (North, three regions of New Jersey of eastern Pennsylvania (Allentown-Easton Central, and South) and two regions and the northern These dummies control for any region- suburbs of Philadelphia). specific tfect of the minimum demand shocks and identify he ef- employment Jersey. stores changes at higher- and wage by comparing no evidence The probability within the same region lower- wage value in row 6o shows f New The addition of the re- iployment growth. of regional components n em- gion dummies attenuates cient and raises the GAP coeffi- it no its standard error, however, making longer possible to reject the null hypothesis of the minimum wage. One explanation of a zero employment effect this attenuation ment error in the starting is the presence employment growth wage. Even if of measure- for nent, the addition of region dummies has no regional compo- lead to some attenuation wGAP coefficient tion in GAP is explained if some of the true varia- of the estimated ill calculations based ity of the GAP variable on the estimated by region. double interviews) (from trIndeed, eliabil- the estimated suggest that the fall he set of 11 (iv) to column (v) is just equal to GAP coefficient from column in ttributable to measurement the ex- pected change aerror.'6 Table 4 using as a dependent We have also estimated the models in proportional change store.'7 The estimated coefficients in employment at each variable the New Jersey dummy of the are uniformly insignificantly different from 0 positive in these models and the GAP variable tional levels. The implied employment at conven- but fects of the minimum wage ef- when the dependent proportional terms. coefficient For example, variable are also smaller is expressed the GAP in that the increase in column (iii) of Table 4 implies employment at New Jersey in minimum wages raised stores that were initially paying $4.25 per hour by 14 per- cent. The estimated from acorresponding an effect proportional GAP coefficient model implies attributable to heterogeneity in the effect of of only 7 percent. The difference is the minimum wage at larger and smaller stores. Weighted versions of tional-change ment as a weight) models (using initial employ- the propor- give rise to wage elastici- 16In texpected attenuation a regression model measurement error of the GAP coefficient without other controls he which of GAP due to factor wwhen e estimate at 0.70. The expected is the reliability ratio (yo), region dummies are added to the model attenuation is statistic of a regression of GAP lYi = (yo - R2)/(1- R2), where R2 to 0.30). Thus, we expect the ois the R-square effects (equal cient to fall by a factor of estimated n region GAP coeffi- dummies are added to a regression model. y1 /yo = 0.8 when region 17These Card and Krueger (1993). specifications are reported in table 4 of 782 THE AMERICAN ECONOMIC REVIEW SEPTEMBER 1994 ties similar to the elasticities implied by the little effect on the models for the level of estimates in Table 4 (see below). C. Specification Tests contradict the standard prediction that a The results in Tables 3 and 4 seem to rise in the minimum ployment. wage will reduce em- specifications Table 5 presents some alternative this conclusion. For completeness, we re- that probe the robustness of port estimates employment of models for the change in mates in employment [columns (iii) and of models [columns for the proportional change (i) and (ii)] and esti- first row of the table reproduces (iv)].'8 The specification\" from columns tTable 4. (Note that these models include (ii) and (iv) of he \"base chain dummies owned stores). Row 2 presents an alterna- and a dummy for company- tive set of estimates when we set wave-2 employment at the temporarily closed to 0 (expanding our sample stores change coefficient has a small size by 4). This all four stores are in New Jersey) of the New Jersey attenuating effect dummy o(n the since effect on the GAP coefficient of GAP is uncorrelated with the probability (since bthe size ut less of a temporary closure within ing alternative Rows 3-5 present estimation New Jersey). results us- alent employment. measures redefined to exclude management In row 3, employment is of full-time-equiv- ees. This change has no effect relative to employ- the base specification. include managers reweight in FTE In rows 4 and employment 5, we cent or 60 percent of full-time part-time workers as either but stead of 50 percent).'9 These changes workers 40 per- h(ave in- 18The fined as the change in employment proportional change in employment is de- average results level of employment in waves 1 and 2. This divided by the errors ployment. tin very han the alternative of dividing similar coefficients but smaller bstandard change in employment to -1. For closed stores we set the proportional y wave-1 em- 1Analysis reveals of the 1991 Current Ptry workers. Katz work that part-time workers in the restaurant indus- opulation Survey about 46 percent as many of part-time workers' hours and Krueger (1992) report hours tas full-time in the fast-food industry is 0.57. to full-time workers' hours hat the ratio employment but yield slightly estimates in the proportional-employment- smaller point change models. from In row 6 we present estimates obtained towns along the New Jersey a subsample that excludes 35 stores iclusion of these stores, which may have a shore. The ex- n different in our sample, seasonal mum-wage effects. leads to slightly pattern than other stores larger in row 7 when we add a set of dummy A similar finding emerges mini- variables that indicate the week of the wave-2 to obtain responses As noted earlier, interview.20 win the first from e made an extra New Jersey setores ffort of stores called three or more times to ob- wave of our survey. The fraction tain an interview than in Pennsylvania. To check was higher in New Jersey ity of our results to this sampling the sensitiv- we reestimated feature, that excludes any stores that were called our models on a subsample back are very more than similar ttwice. The results, in row 8, sults for the proportional-employment- Row 9 presents weighted estimation o the base specification. re- change models, using as weights the initial levels of employment the proportional change in each store. Since ment is an employment-weighted in average employ- the proportional weighted changes at each store, a average of model should give rise to elasticities that version of the proportional-change are similar from to the implied elasticities expectation, the weighted estimates are the levels models. Consistent with this arising larger than the unweighted significantly different levels. The weighted estimate of the New from 0 at conventional estimates, and Jersey dummy relative (0.13) implies a 13-percent -the in New Jersey implied by the simple difference-in-dif- same proportional employment effect increase employment ferences in Table 3. estimate of the GAP coefficient in the Similarly, the weighted proportional-change model (0.81) is close to for the wave-1 20We also added dummies for the interview did not change sturvey, but these were insignificant dates he estimated minimum-wage effects. and VOL. 84 NO. 4 CARD AND KRUEGER: MINIMUM WAGE AND EMPLOYMENT TABLE 5-SPECIFICATION TESTS OF REDUCED-FORM EMPLOYMENT MODELS 783 Specification 1. Base specification 2. Treat four temporarily closed stores as permanently closeda 3. Exclude managers in employment countb 4. Weight part-time as 0.4 x full-timec 5. Weight part-time as 0.6 x full-timed 6. Exclude stores in NJ shore areae 7. Add controls for wave-2 interview datef 8. Exclude stores called more than twice in wave 19 9. Weight by initial employmenth 10. Stores in towns around Newark' 11. Stores in towns around Camdeni 12. Pennsylvania stores onlyk Change in employment NJ dummy Gap measure (i) (ii) 2.30 (1.19) 2.20 (1.21) 2.34 (1.17) 2.34 (1.20) 2.27 (1.21) 2.58 (1.19) 2.27 (1.20) 2.41 (1.28) 14.92 (6.21) 14.42 (6.31) 14.69 (6.05) 15.23 (6.23) 14.60 (6.26) 16.88 (6.36) 15.79 (6.24) 14.08 (7.11) Proportional change in employment NJ dummy Gap measure (iv) (iii) 0.05 (0.05) 0.04 (0.05) 0.05 (0.07) 0.06 (0.06) 0.04 (0.06) 0.06 (0.05) 0.05 (0.05) 0.05 (0.05) 0.13 (0.05) 0.34 (0.26) 0.34 (0.27) 0.28 (0.34) 0.30 (0.33) 0.17 (0.29) 0.42 (0.27) 0.40 (0.26) 0.31 (0.29) 0.81 (0.26) 0.90 (0.74) 0.21 (0.70) -0.33 (0.74) 33.75 (16.75) 10.91 (14.09) -0.30 (22.00) - Notes: Standard errors are given in parentheses. Entries represent estimated coefficient of New Jersey dummy [columns (i) and (iii)] or initial wage gap [columns (ii) and (iv)] in regression models for the change in employment or the percentage change in employment. All models also include chain dummies and an indicator for company- owned stores. aWave-2 employment at four temporarily closed stores is set to 0 (rather than missing). bFull-time equivalent employment excludes managers and assistant managers. CFull-time equivalent employment equals number of managers, assistant managers, and full-time nonmanage- ment workers, plus 0.4 times the number of part-time nonmanagement workers. dFull-time equivalent employment equals number of managers, assistant managers, and full-time nonmanage- ment workers, plus 0.6 times the number of part-time nonmanagement workers. eSample excludes 35 stores located in towns along the New Jersey shore. fModels include three dummy variables identifying week of wave-2 interview in November-December 1992. gSample excludes 70 stores (69 in New Jersey) that were contacted three or more times before obtaining the wave-1 interview. hRegression model is estimated by weighted least squares, using employment in wave 1 as a weight. 'Subsample of 51 stores in towns around Newark. J Subsample of 54 stores in town around Camden. k Subsample of Pennsylvania stores only. Wage gap is defined as percentage increase in starting wage necessary to raise starting wage to $5.05. 784 THE AMERICAN ECONOMIC REVIEW SEPTEMBER 1994 the implied elasticity of employment respect fication to wages from the basic levels speci- with ings suggest that the proportional in row 1, column (u).21 These find- the rise in the minimum wage was concen- effect of trated among lrise in the minimum One explanation arger sfor our finding that a tores. employment effect is that unobserved wage has a positive mand de- the negative shocks mum ewmployment effect of the ithin New Jersey outweighed 10 and 11 present estimation wage. To address this possibility, rows mini- on subsamples of stores in two narrowly results based defined areas: towns around Newark (row 10) and towns around each case the sample area is identified Camden (row 11). In the first Within both areas the change in employ- three digits of the store's zip code.22 by ment is positively variable, although in neither case is the correlated with the GAP effect statistically that constant within local areas, these results fast-food product significant. To the extent market conditions are suggest that our findings unobserved demand price changes (reported below) also sup- shocks. are not driven Our analysis boy f ports row 12 of Table 5. In this row we exclude A final this conclusion. specification check is presented in stores in New Jersey and (incorrectly) de- fine the GAP variable for Pennsylvania stores as the proportional increase necessary to raise the wage to $5.05 per in wages hour. In principle for stores in Pennsylvania should have no the size of the wage gap systematic relation In practice, this is the case. There is no with employment growth. indication that the wage gap is spuriously related to employment growth. Jersey, 2'Assuming average employment implies tahe 14.92 n employment elasticity of 0.73. GAP coefficient in row 1, column of 20.4 in New (ii) 2The Newark) and the \"080\" three-digit zip-code area \"070\" three-digit zip-code area (around (around stores Camden) have by far the largest numbers of and among three-digit zip-code areas in New Jersey, stores together in our sample. they account for 36 percent of New Jersey first-differenced We have also investigated employment models is appropriate. A specification used in our whether the first-differenced of employment in period t is related model implies that the level lagged level of employment cient of 1. If short-run employment fluctua- with a coeffi- to the tions are smoothed, however, efficient tthan 1. Imposing of lagged employment he true co- may coefficient of a unit be less the first-differenced specification we reesti- may then lead to biases. To test the assumption mated models for the change in employ- ment including additional explanatory wave-1 employment come any mechanical base-period cvorrelation ariable. To over- as an between employment employment and the change in error) with the number of cash registers in the we instrumented wave-1 (attributable to measurement employment store in wave 1 and the number in the store that of registers A.M. all of the specifications were open at 11:00 wave-1 employment the coefficient of In example, in a specification including the is close to zero. For GAP variable and ownership and chain dummies, ment is 0.04, with a standard the coefficient of wave-1 We conclude error employ- ification is appropriate. that the first-differenced of 0.24. spec- D. Full-Time and Part-Time Substitution full-time-equivalent employment Our analysis so far has concentrated on nored possible changes in the distribution and ig- of full- and part-time in the minimum An increase crease in full-time employment wage could lead to an in- workers. part-time sons. First, in a conventional model one employment for at least two rea- relative to would expect a minimum-wage increase to induce ers and capital employers to substitute skilled work- Full-time workers in fast-food restaurants for minimum-wage workers. are typically older and may well possess higher conventional model predicts skills than part-time workers. Thus, respond to an increase in the minimum that stores may a wage by increasing time workers. Nevertheless, 81 percent of the proportion of full- restaurants paid full-time and part-time VOL. 84 NO. 4 CARD AND KRUEGER: MINIMUM WAGE AND EMPLOYMENT 785 TABLE 6-EFFECTS OF MINIMUM-WAGE INCREASE ON OTHER OUTCOMES Regression of change in Mean change in outcome outcome variable on: NJ PA NJ - PA NJ dummy Wage gapa Wage gapb Outcome measure (i) (ii) (iii) (iv) (v) (vi) Store Characteristics: 1. Fraction full-time workersc (percentage) 2.64 -4.65 7.29 7.30 33.64 20.28 (1.71) (3.80) (4.17) (3.96) (20.95) (24.34) 2. Number of hours open per weekday -0.00 0.11 -0.11 -0.11 -0.24 0.04 (0.06) (0.08) (0.10) (0.12) (0.65) (0.76) 3. Number of cash registers - 0.04 0.13 -0.17 -0.18 -0.31 0.29 (0.04) (0.10) (0.11) (0.10) (0.53) (0.62) 4. Number of cash registers open -0.03 -0.20 0.17 0.17 0.15 -0.47 at 11:00 A.M. (0.05) (0.08) (0.10) (0.12) (0.62) (0.74) Employee Meal Programs: 5. Low-price meal program (percentage) - 4.67 - 1.28 - 3.39 - 2.01 -30.31 - 33.15 (2.65) (3.86) (4.68) (5.63) (29.80) (35.04) 6. Free meal program (percentage) 8.41 6.41 2.00 0.49 29.90 36.91 (2.17) (3.33) (3.97) (4.50) (23.75) (27.90) 7. Combination of low-price and free -4.04 -5.13 1.09 1.20 -11.87 -19.19 meals (percentage) (1.98) (3.11) (3.69) (4.32) (22.87) (26.81) Wage Profile: 8. Time to first raise (weeks) 3.77 1.26 2.51 2.21 4.02 -5.10 (0.89) (1.97) (2.16) (2.03) (10.81) (12.74) 9. Usual amount of first raise (cents) -0.01 -0.02 0.01 0.01 0.03 0.03 (0.01) (0.02) (0.02) (0.02) (0.11) (0.11) 10. Slope of wage profile (percent -0.10 -0.11 0.01 0.01 -0.09 -0.08 per week) (0.04) (0.09) (0.10) (0.10) (0.56) (0.57) Notes: Entries in columns (i) and (ii) represent mean changes in the outcome variable indicated by the row heading for stores with available data on the outcome in waves 1 and 2. Entries in columns (iv)-(vi) represent estimated regression coefficients of indicated variable (NJ dummy or initial wage gap) in models for the change in the outcome variable. Regression models include chain dummies and an indicator for company-owned stores. aThe wage gap is the proportional increase in starting wage necessary to raise the wage to the new minimum rate. For stores in Pennsylvania, the wage gap is zero. bModels in column (vi) include dummies for two regions of New Jersey and two regions of eastern Pennsylvania. CFraction of part-time employees in total full-time-equivalent employment. workers wave 1 of our survey.23 exactly the same starting wage in workers are more productive that full-time paid), there may be a second reason for (but equally as part-time workers or that equity workers hThis suggests ave the same skills either stores to substitute full-time workers for lead restaurants to pay equal concerns part-time workers; namely, equally productive workers. If full-time wages for un- increase a minimum-wage full-time workers, and stores would natu- enables the industry to attract more rally want to hire a greater proportion full-time tive. workers if they are more produc- of 231n the other 19 percent of stores, full-time workers are paid more, typically 10 percent more. changes Row 1 of Table 6 presents the mean in the proportion of full-time work- 786 THE AMERICAN ECONOMIC REVIEW SEPTEMBER 1994 ers in New Jersey and Pennsylvania coefficient 1 and 2 of our survey, and be- tween waves change in the proportion estimates from regressions of the ers on the wage-gap mies, a company-ownership variable, chain dum- of full-time work- gion dummies dummy, and re- are ambiguous. The fraction of full-time [in column (vi)]. The results workers Pennsylvania by 7.3 percent increased in New Jersey relative = 1.84), to but regressions on the wage-gap variable (t ratio show no significant full-time workers.24 shift in the fraction of E. Other Employment-Related Measures other outcome variables Rows 2-4 of Table 6 present results for be related that we expect to ment. In particular, to the level of restaurant employ- the rise in the minimum wage is associated we examine whether with a change in the number of hours a restaurant is open on a weekday, ber of cash registers tthe number of cash registers typically in the restaurant, he num- and at 11:00 in operation in the restaurant A.M. Consistent with our employment results, none of these variables significant decline shows a statistically Pennsylvania. Similarly, in New Jersey relative ing the gap variable provide no evidence regressions includ- to that the minimum-wage increase led to a systematic [see columns change in any of these variables (v) and (vi)]. IV. Nonwage Offsets in the minimum One explanation of our finding ployment effect of the minimum is that restaurants wage does not lower em- that a rise can offset the nonwage compensation. wage by reducing workers value fringe benefits and wages For example, if equally, employers can simply reduce level of fringe the minimum-wage increase, leaving benefits by the amount their em- of the ployees 24Within New Jersey, the fraction of full-time eand lower increased about wages in wave 1. as quickly at stores with higher m- ployment benefits for fast-food employees are free costs unchanged. The main fringe and reduced-price meals. of our survey restaurants about 19 percent In the first wave opercent offered reduced-price meals, offered workers free meals, 72 f fast-food percent offered a combination and reduced-price of both free and 9 are an obvious fringe benefit to cut if the meals. Low-price meals minimum-wage increase forces restaurants to pay higher wages. mates of the effect of the minimum-wage Rows 5 and 6 of Table 6 present esti- increase reduced-price meals. on the incidence taurants offering reduced-price The proportion of res- of free meals and in both New Jersey and Pennsylvania after meals fell the minimum wage increased, what greater decline in New Jersey. Con- with a some- trary tion in reduced-price to an offset story, however, the reduc- accompanied by an meal programs was of stores offering free meals. Relative to increase in the fraction stores in Pennsylvania, New Jersey ers actually employ- fringe benefits (i.e., free meals rather shifted toward more generous reduced-price meals). shift is not statistically significant. However, the relative than We continue to find a statistically isignificant crease effect of the minimum-wage in- n- reduced-price meals on the likelihood where in columns of receiving free or (v) and (vi), GAP variable from regression models we report coefficient estimates of the the change in the incidence of these pro- for grams. temployers offset The results pthe minimum-wage rovide no evidence hat in- crease by reducing free or reduced-price meals. is that employers re- sponded to Another possibility wage by reducing on-the-job training the increase in the minimum and flattening Jacob Mincer and Linda Leighton, 1981). the tenure-wage profile (see Indeed, in wave 1 that her workers were forgoing one manager told our interviewer ordi- nary wage was about scheduled raises provide to because the minimum termine arise, and this would more generally, we analyzed store man- w raise for all her workers. hether this phenomenon occurred To de- agers' responses to questions on the amount VOL. 84 NO. 4 AND EMPLOYMENT CARD AND KRUEGER: MINIMUM WAGE 787 of time before a normal the usual amount of such raises. In rows wage increase and 8 and 9 we report the average tween as well as regression waves 1 and 2 for these two changes be- els that include coefficients from variables, mod- though the average time to the the wage-gap variable.25 Al- raise increased relative statistically to Pennsylvania, the increase by 2.5 weeks in New Jersey first pay is not is only a trivial difference significant. Furthermore, there change ment between in the amount of the first in the relative pay incre- stores. New Jersey and Pennsylvania Finally, we examined a related variable: the \"slope\" of the wage profile, measure to the amount by the ratio which we given. oof the typical first raise wage profile flattened in both As shown f time until the first in row 10, the slope of the raise is New Jersey and Pennsylvania, with no significant tive difference rin the slope is also uncorrelated between states. The change ela- GAP variable. indication that New Jersey In summary, we can find no with the changed wage profiles either their fringe mum wage.26 to offset the rise benefits employers in the mini- or their V. Price Effects of the Minimum-Wage Increase the minimum wage A final issue we examine fast-food restaurants. on the prices is the effect of of meals mat of the fast-food industry implies A competitive that an odel increase an increase in the minimum wage will lead to constant increase returns in product to scale in the industry, the prices. If we assume the share of minimum-wage in price should be proportional to labor in total 25In wave 1, the average time to a first wage in- crease was 18.9 weeks, and the average amount of the first increase was $0.21 per hour. 26Katz and Krueger (1992) report that a significant fraction of fast-food stores in Texas responded to an increase in the minimum wage by raising wages for workers who were initially earning more than the new minimum rate. Our results on the slope of the tenure profile are consistent with their findings. factor cost. The average Jersey initially restaurant in New less than the new minimum paid about half its workers rose by roughly ers, and if labor's 15 percent for these work- wage. If wages percent, we would expect share of total costs is 30 prices to rise by the minimum-wage rise.27 about 2.2 percent (= 0.15 x 0.5 x 0.3) due to managers In each wave of our survey we asked items: a medium soda, for the prices of three standard french fries, and a main course. a small order of The main course was a basic hamburger at Burger King, and two pieces of chicken at KFC Roy Rogers, and Wendy's restaurants, We define \"full price of a medium soda, meal\" price as the after-tax stores. french fries, and a main course. a small order of of the effect of the minimum-wage Table 7 presents reduced-form estimates on prices. The dependent models in the vincrease in these price of a full is the change logarithm ariable of the independent dicating whether variable meal at each store. The key Jersey or the proportional the store is either a dummy is located in New in- required GAP variable to meet the minimum wage increase wage (the umn (i) shows that after-tax The estimated defined New Jersey above). dummy in col- rose 3.2-percent in Pennsylvania between February faster in New Jersey meal prices than and November larger controlling 1992.28 The effect is slightly ownership [see New Jersey sales tax rate fell by column (ii)]. Since for chain and company- 1 the age point between these estimates suggest the waves of our survey, percent- that rose 4-percent faster as a result of the pretax prices 27According to the McDonald's operating costs at company-owned and aCre 31.3 orporation percent 1991 Annual Report, payroll benefits This cof tion is only ers make approximate because mstores. inimum-wage alcula- ework- they are about half of up less than half of payroll ven though minimum wage causes workers, and because a rise in the the pay relative of other some employers to increase pay differentials. higher-wage workers in order to maintain in prices 28The effect is attributable to a 2.0-percent increase prices in Pennsylvania. in New Jersey and a 1.0-percent decrease in 788 THE AMERICAN ECONOMIC REVIEW TABLE 7-REDUCED-FORM SEPTEMBER 1994 MODELS FOR CHANGE IN THE PRICE OF A FULL MEAL Dependent variable: change in the log price of a full meal Independent variable 1. New Jersey dummy 2. Initial wage gapa 3. Controls for chain andb ownership 4. Controls for regionc 5. Standard error of regression no no 0.101 (i) 0.033 (0.014) (ii) 0.037 (0.014) - (iii) (iv) (v) 0.077 (0.075) no no 0.102 0.146 (0.074) yes no 0.098 0.063 (0.089) yes yes 0.097 yes no 0.097 Notes: Standard errors are given in parentheses. Entries are estimated regression coefficients for models fit to the change in the log price of a full meal (entree, medium soda, small fries). The sample contains 315 stores with valid data on prices, wages, and employment for waves 1 and 2. The mean and standard deviation of the dependent variable are 0.0173 and 0.1017, respectively. aProportional increase in starting wage necessary to raise the wage to the new minimum-wage rate. For stores in Pennsylvania the wage gap is 0. bThree dummy variables for chain type and whether or not the store is company- owned are included. CDummy variables for two regions of New Jersey and two regions of eastern Pennsylvania are included. minimum-wage increase in New Jersey- slightly more than the increase needed to pass through the cost increase caused by the minimum-wage hike. The pattern of price changes within New Jersey is less consistent with a simple \"pass-through\" view of minimum-wage cost increases. In fact, meal prices rose at approximately the same rate at stores in New Jersey with differing levels of initial wages. Inspection of the estimated GAP coefficients in column (v) of Table 7 con- firms that within regions of New Jersey, the GAP variable is statistically insignificant. In sum, these results provide mixed evi- dence that higher minimum wages result in higher fast-food prices. The strongest evi- of New dence emerges from a comparison Jersey and Pennsylvania stores. The magni- is consistent with tude of the price increase from a conventional model of a predictions we competitive industry. On the other hand, find no evidence that prices rose faster among stores in New Jersey that were most affected by the rise in the minimum wage. for the latter One potential explanation compete finding is that stores in New Jersey in the same product market. As a result, restaurants that are most affected by the minimum wage are unable to increase their faster than their competitors. product prices and Penn- In contrast, stores in New Jersey sylvania are in separate product markets, enabling prices to rise in New Jersey rela- tive to Pennsylvania when overall costs rise in New Jersey. Note that this explanation that store- seems to rule out the possibility for the demand shocks can account specific at stores in anomalous rise in employment with lower initial wages. New Jersey VI. Store Openings An important potential effect of higher the open- minimum wages is to discourage our sample Although ing of new businesses. the effect of the us to estimate design allows in minimum wage on existing restaurants the effect of we cannot address New Jersey, the higher minimum wage on potential VOL. 84 NO. 4 CARD AND KRUEGER: MINIMUM WAGE AND EMPLOYMENT 789 entrants.29 To assess the likely size of such an effect, tories for the McDonald's we used national rto compare the numbers of operating restaurant direc- estaurant chain restaurants and the numbers of newly opened restaurants the 1986-1991 period. Many states raised in different states over their minimum wages during addition, the federal minimum wage in- this period. In creased in the early 1990's from $3.35 to $4.25, depending on the level of wages in the with differing effects in different states state. These policies create an opportunity to measure the impact of minimum-wage laws on store opening rin Table 8. We regressed The results of our analysis ates across are presented states. the number of McDonald's the growth rate in state on two alternative minimum measures of the stores in each other control variables wage in the state and a set of and the change in the state unemployment (population growth rate). The first minimum-wage measure is the fraction trade industry in 1986 of workers in the state's retail tween the existing whose wages fell be- 1986 imum ($3.35 per hour) federal and the effective minimum wage in min- maximum of the federal wage in the state in April 1990 (the the state minimum wage and The second is the ratio of the state's effec- minimum wages as of April 1990).30 tive minimum hourly wage in 1990 to the average state in 1986. Both of these measures are wage of retail trade workers in the designed to gauge the degree of upward wage pressure exerted by state or federal minimum-wage changes between 1986 and 1990. higher The results provide no evidence that retail-trade wages in a state) exert a nega- minimum-wage rates (relative to the vealed that Wendy's opened two stores 29Direct inquiries to the chains in our sample re- in 1992 and one store in Pennsylvania. The other in New Jersey chains were unwilling to provide information on new openings. (merged 30We used the 1986 Current Population Survey variables. State minimum-wage monthly file) to construct the minimum-wage tained from the Bureau of National Affairs Labor rates in 1990 were ob- Relations Reporter Wages and Hours Manual (undated). tive effect on either the net number of restaurants or the rate of new openings. the contrary, all the estimates To effects of higher minimum number of operating or newly wshow positive although many of the point estimates are opened ages on the stores, insignificantly evidence is limited, we conclude that the different from zero. While this effects opening of minimum wages rates are probably small. on fast-food store VII. Broader Evidence on Employment Changes in New Jersey that the rise in the minimum Our establishment-level analysis suggests Jersey may have increased employment wage in New the fast-food iassociated phenomenon with our particular sample, or a industry. Is this just an anomaly n try? Data from the monthly unique to the fast-food lation Survey (CPS) allow us to compare Current iPndus- opu- state-wide sey and the surrounding states, providing employment trends in New Jer- check on the interpretation of our findings. a Using monthly we computed CPS files for 1991 and 1992, for teenagers efor New Jersey, Pennsylvania, amployment-population rates nd adults (age 25 and older) and the entire Jersey United States. Since the New New York, we computed the employment rates for minimum wage rose on April 1, 1992, April-December of both 1991 The relative and 1992. Jersey changes in employment in New an indication and the ployment rates show that the New Jersey A comparison of the effect of the new law. surrounding states then give of changes in adult em- labor market fared slightly market as a whole or labor markets in 1991-1992 period than either wtorse over the he U.S. labor Pennsylvania or New York (see Card Krueger, however, the situation 1993 table and 9).31 Among teenagers, Jersey, was reversed. In New percent teenage employment rates from 1991 to 1992. In New fell by 0.7 York, older fell by 2.6 percent 31The employment rate of individuals age 25 and and 1992, in New Jersey and fell by 0.2 percent while it rose by 0.3 percent in the United Sitates n Pennsylvania, between 1991 as a whole. 790 TABLE 8-ESTIMATED THE AMERICAN ECONOMIC REVIEW EFFECT OF MINIMUM WAGES ON NUMBERS OF MCDONALD'S SEPTEMBER 1994 RESTAURANTS, 1986-1991 Independent variable Variable: Minimum-Wage 1. Fraction of retail workers Dependent variable: proportional increase in number of stores (ii) (iv) (i) (iii) Dependent variable: + (number of newly opened stores) (number in 1986) (v) (vi) (vii) (viii) in affected wage 1986a range 2. (State minimum wage in 1991) Other Control Variables: 3. Proportional growth in (0.20) - 0.33 (0.19) (0.22) 0.38 0.13 - (0.22) (0.22) 0.47 0.37 - (0.21) (0.23) 0.47 - 0.16 in 1986)b (average retail wage (0.24) 0.56 - - 4. Change in unemployment population, 1986-1991 rates, 1986-1991 of regression 5. Standard error - -1.78 (0.23) (0.62) 0.071 0.88 - 1.40 (0.23) (0.61) 0.068 0.088 1.03 - - - 1.85 (0.25) (0.68) 0.077 0.86 - 1.40 (0.25) (0.65) 0.073 1.04 0.083 0.083 0.088 Notes: Standard errors are shown in parentheses. The sample contains 51 state-level observations (including the District of Columbia) on the number of McDonald's restaurants open in 1986 and 1991. The dependent variable in (i)-(iv) is the proportional increase columns in the number of restaurants open. The mean and standard deviation are 0.246 and 0.085, respectively. The dependent variable in columns (v)-(viii) is the ratio of the number of new stores between 1986 opened and 1991 to the number open in 1986. The mean and standard deviation are 0.293 and 0.091, respectively. All regressions are weighted by the state population in 1986. aFraction of all workers in retail trade in the state in 1986 earning an hourly wage between $3.35 per hour and the \"effective\" state minimum wage in 1990 (i.e., the maximum of the federal minimum wage in 1990 ($3.80) and the state minimum wage as of April 1, 1990). bMaximum of state and federal minimum wage as of April 1, 1990, divided hourly wage of by the average workers in retail trade in the state in 1986. and the United States as a Pennsylvania, rates dropped whole, teenage employment faster. Relative to teenagers in Pennsylva- nia, for example, the teenage employment rate in New Jersey rose by 2.0 percentage points. While this point estimate is consis- for the fast-food in- tent with our findings error is too large (3.2 dustry, the standard assessment. percent) to allow any confident of the standard the predictions summarize model and some simple alternatives, and we posed by our find- the difficulties highlight ings. Model A. Standard Competitive A standard competitive model predicts employment will fall that establishment-level raised. For an if the wage is exogenously entire industry, total employment is pre- VIII. Interpretation price is predicted dicted to fall, and product in a bind- to an increase As in the earlier study by Katz and to rise in response Krueger (1992), our empirical findings on ing minimum wage. Estimates from the on minimum-wage ef- literature minimum wage time-series of the New Jersey the effects are inconsistent with the predictions of a fects can be used to get a rough idea of the model of the fast- elasticity of low-wage employment to the competitive conventional et al. results are minimum wage. The surveys by Brown Our employment food industry. in- models, (1982, 1983) conclude that a 10-percent consistent with several alternative minimum although none of these models can also crease in the coverage-adjusted rise in fast-food prices wage will reduce teenage employment rates the apparent explain in New Jersey. In this section we briefly by 1-3 percent. Since this effect is for all VOL. 84 NO. 4 CARD AND KRUEGER: MINIMUM WAGE AND EMPLOYMENT 791 teenagers, and not just those employed in low-wage industries, it is surely a lower bound on the magnitude fast-food workers. in the New Jersey The 18-percent of the effect for increase fore predicted to reduce employment at minimum wage is there- fast-food stores by 0.4-1.0 employees per store. Our empirical results clearly reject the upper range of these estimates, al- though we cannot reject a small negative effect in some of our specifications. model is that unobserved A possible defense of the competitive affected certain stores in New Jersey- demand shocks specifically, paying wages those stores that were initially ever, such localized demand shocks should less than $5.00 per hour. How- also affect product prices. (In fact, in a competitive work through a rise in prices.) Although model, product demand shocks lower-wage tive employment stores in New Jersey had rela- relative price increases. Furthermore, gains, they did not have jor suburban areas (around Newark and analysis of employment changes in two ma- our Camden) reveals that, even within local areas, employment rose faster at the stores that had to increase of the new minimum wage. wages the most because B. Altemative Models petitive model is one An alternative to the conventional price-takers in the product in which firms are com- some degree of market market. power market in the labor but have sloping labor-supply schedule, a rise in the If fast-food stores face an upward- minimum ployment wage can potentially increase etry at affected firms and in the indus- m- equilibrium This same basic insight as a whole.32 emerges from an post wages and employees search among search model in which firms posted offers Kenneth Burdett (see Dale T. Mortensen, and Mortensen (1989) 1988). de- (1993) 32Daniel G. Sullivan (1989) and Msity teachers present empirical results for nurses ichael R Ransom employers. that suggest monopsony-like behavior and univer- of rive the equilibrium wage distribution for a noncooperative wage-search/wage-posting model and show that the imposition of a binding minimum wages equilibrium. Furthermore, their model pre- and employment relative wage can increase both to the initial dicts that the minimum employment wage will increase paid the lowest the most at firms that initially models provide Although monopsonistic and job-search wages. the observed employment effects a potential explanation for Jersey of the New the observed minimum industry Jersey prices should have fallen in New price effects. In these models, wage, they cannot explain wage stores in New Jersey relative to Pennsylvania, and at low- wage stores in New Jersey. rtion is confirmed: indeed, prices Nelative either predic- to high- in New Jersey than in Pennsylvania, rose faster though al- low-wage stores in New Jersey. Another at about the same rate at high- and puzzle for equilibrium search absence initially of wage increases at firms models is the that The strict link between the employment paying $5.05 or more per hour. were wage may and price effects of a rise in the minimum vary the queue at peak hours, the quality be broken of service if fast-food (stores can of stores). altered the Another possibility oe.g., the length r the cleanliness of is that stores menu items. Comparisons of price changes relative prices of their various for the three items in our declines (- 1.5 percent) in the price of survey show slight french to Pennsylvania, fries and soda in New Jersey crease (8 percent) in entree prices. These coupled with a relative relative in- limited tive price changes within the fast-food in- data suggest a possible role for rela- dustry following the rise in the minimum wage. identify stores that were initially \"supply- One way to test a monopsony model is to constrained\" for in the labor market and test tive to other stores. employment gains at these stores rela- market power is the use of recruitment A potential indicator of bonuses. As we noted in Table percent 2, about 25 cash bonuses of stores in wave 1 were a new worker. We to employees offering compared who helped find employment 792 THE AMERICAN ECONOMIC REVIEW SEPTEMBER 1994 changes at New Jersey stores that were of- bonuses in wave 1, and fering recruitment also interacted the GAP variable with a dummy for recruitment bonuses in several models. We do not find employment-change faster (or slower) employment growth at the New Jersey stores that were initially using recruitment bonuses, or any evidence that the GAP variable had a larger effect for stores that were using bonuses. IX. Conclusions affected by the minimum-wage rise. Taken as a whole, these findings are difficult to with the standard competitive model explain or with models in which employers face supply constraints (e.g., monopsony or equi- librium search models). REFERENCES Brown, Charles; Gilroy, Curtis and Kohen, An- of the Contrary to the central prediction textbook model of the minimum wage, but of recent studies consistent with a number based on cross-sectional time-series com- of affected and unaffected markets parisons or employers, we find no evidence that the rise in New Jersey's minimum wage reduced at fast-food restaurants in the employment state. Regardless of whether we compare stores in New Jersey that were affected by the $5.05 minimum to stores in eastern Pennsylvania (where the minimum wage was constant at $4.25 per hour) or to stores in New Jersey that were initially paying $5.00 per hour or more (and were largely unaf- fected by the new law), we find that the increase in the minimum wage increased We present a wide variety of employment. alternative specifications to probe the ro- bustness of this conclusion. None of the shows a negative employment alternatives effect. We also check our findings for the fast-food industry by comparing changes in teenage employment rates in New Jersey, and New York in the year Pennsylvania, in the minimum the increase following wage. a relative Again, these results point toward increase in employment of low-wage work- ers in New Jersey. We also find no evidence that minimum-wage increases negatively affect the number of McDonald's outlets opened in a state. Finally, we find that prices of fast-food meals increased in New Jersey relative to Pennsylvania, suggesting that much of the burden of the minimum-wage rise was Within New Jer- passed on to consumers. that sey, however, we find no evidence prices increased more in stores that were most drew. \"The Effect of the Minimum Wage on Employment and Unemployment.\" Journal of Economic Literature, June 1982, 20(2), pp. 487-528. \"Time Series Evidence on the Ef- fect of the Minimum Wage on Youth Employment and Unemployment.\" Jour- nal of Human Resources, Winter 1983, 18(1), pp. 3-31. _ Burdett, Kenneth and Mortensen, Dale T. \"Equilibrium Wage Differentials and Employer Size.\" Center for Mathematical Studies in Economics and Management Science Discussion Paper No. 860, North- western University, October 1989. Bureau of National Affairs. Daily Labor Re- port. Washington, DC: Bureau of Na- tional Affairs, 5 May 1990. and . Labor Relations Reporter Wages Hours Manual. Washington, DC: Bureau of National Affairs, irregular. Card, David. \"Using Regional Variation in Wages To Measure the Effects of the Federal Minimum Wage.\" Industrial and Labor Relations Review, October 1992a, 46(1), pp. 22-37. . \"Do Minimum Wages Reduce Em- ployment? A Case Study of California, 1987-89.\" Industrial and Labor Relations Review, October 1992b, 46(1), pp. 38-54. Card, David and Krueger, Alan B. \"Minimum Wages and Employment: A Case Study of the Fast Food Industry in New Jersey and Pennsylvania.\" National Bureau of Eco- nomic Research (Cambridge, MA) Work- ing Paper No. 4509, October 1993. Katz, Lawrence F. and Krueger, Alan B. \"The Effect of the Minimum Wage on the Fast Food Industry.\" Industrial and Labor Re- lations Review, October 1992, 46(1), pp. 6-21. VOL. 84 NO. 4 CARD AND KRUEGER: MINIMUM WAGE AND EMPLOYMENT 793 Lester, Richard A. \"Employment Effects of Minimum Wages.\" Industrial and Labor Relations Review, January 1960, 13, pp. 254-64. . The economics of labor, 2nd Ed. New York: Macmillan, 1964. Machin, Stephen and Manning, Alan. \"The Effects of Minimum Wages on Wage Dis- persion and Employment: Evidence from the U.K. Wage Councils.\" Industrial and Labor Relations Review, January 1994, 47(2), pp. 319-29. McDonald's Corporation. 1991 Annual report. Chicago, 1991. Mincer, Jacob and Leighton, Linda. \"The Ef- fects of Minimum Wages on Human Cap- ital Formation,\" in Simon Rottenberg, ed., The economics of legal minimum wages. Washington, DC: American Enter- prise Institute, 1981, pp. 155-73. Dale T. \"Equilibrium Wage Dis- Mortensen, tributions: A Synthesis.\" Center for Mathematical Studies in Economics and Management Science Discussion Paper No. 811, Northwestern University, March 1988. Ransom, Michael R. \"Seniority and Monop- sony in the Academic Labor Market.\" American Economic Review, March 1993, 83(1), pp. 221-33. Stigler, George J. \"The Economics of Mini- mum Wage Legislation.\" American Eco- nomic Review, June 1946, 36(3), pp. 358-65. Daniel G. \"Monopsony Power in the Sullivan, Market for Nurses.\" Journal of Law and Economics, October 1989, Part 2, 32(2), pp. S135-78. U.S. Department of Commerce. 1987 Census of retail trade: Miscellaneous subjects. Wash- ington, DC: U.S. Government Printing Office, October 1990. Alison J. \"Effects of the Mini- Wellington, mum Wage on the Employment Status of Youths: An Update.\" Journal of Human Resources, Winter 1991, 26(1), pp. 27-46.