Minimum Wage Policy and Poverty in the United States

ISSN 0269-2171 print/ISSN 1465 -3486 online/01/010065-11 � 2001 Taylor & Francis LtdDOI: 10.1080/026921701 20013358

International Review of Applied Economics, Vol. 15, No. 1, 2001

Lonnie K. Stevans, Department of BCIS/QM, 134 Hofstra University, Hempstead, NY 11549 ± 1340,

USA. E-mail: acslks@hofstra.edu. Web Page: http:\\www.i2.i ± 2000.com \ ~ acslks\lks.html

David N. Sessions, Department of BCIS/QM, 134 Hofstra University, Hempstead, NY 11549 ± 1340,

USA. E-mail: acsdns@hofstra.edu

Minimum Wage Policy and Poverty in the United

States

LONNIE K. STEVANS & DAVID N. SESSIONS

ABSTRACT Recent studies have found that increasing the minimum wage is a useful anti-

poverty tool. In this analysis, we examine the influence of minimum wages and other

important variables on US family poverty rates using state data over the years 1984 ± 98

by estimating both a fixed effect and random coefficients regression model. Taking into

account labor market influences, demographic factors, and differences in poverty rates

across states, we find that expanding the minimum wage coverage and increasing labor

force participation both have larger effects on poverty rates as compared to equivalent

changes in the level of the minimum wage. It is further implied from the empirical results

that the most effective means of lifting families out of poverty are policies that are directed

toward increasing minimum wage coverage, encouraging increased labor force participa-

tion, raising the minimum wage, and subsidizing higher education, respectively.

1. Introduction

Increasing the minimum wage has been viewed as one way of reducing poverty

(Gramlich, 1976; Freeman, 1996), by raising the living standards of low-income

workers and families. However, if increases in the minimum wage reduce

employment, the effects on poverty are uncertain. In analyzing these effects,

researchers have relied on simulation exercises that depend upon a number of

dubious assumptions; such as disemployment effects and certain labor supply

responses. In this study, we do not incorporate such constraints and thus avoid the

problems associated with simulation exercises.

In a study by Card & Krueger (1995), it was found that the 1990 ± 91 US

minimum wage hike led to modestly lower poverty rates. Even & Macpherson

(1996) disputed that minimum wage policies have an effect on poverty rates. These

contradictory results provide little guidance for policy-makers. In addition, the

entrustment of welfare programs to the states by the US federal government

indicates that research on these issues should be targeted at the state-level. In this

66 L. K. Stevans & D. N. Sessions

study, we will examine the effects that minimum wage laws have on US poverty rates

using data that are available by state over the period 1984 to 1998. Because we will

relate poverty to minimum wages, minimum wage coverage, and other factors, there

is no need to incorporate any restrictive assumptions as is done in simulation

studies.1

2. Literature Review

Minimum wage hikes are publicly supported because they provide a `living wage’ .

However, many economists have argued that raising the minimum wage will lead to

declines in employment and reductions in benefits. These losses are presumed to be

concentrated among teens and minorities Ð the very workers that the minimum

wage is supposed to benefit (Brown et al., 1982). Results from recent empirical

studies have not been consistent with these hypotheses. Card (1992), and Card &

Krueger (1995) found little evidence that federal minimum wage increases in 1990

and 1991 reduced employment. In an analysis of the effects of minimum wages on

employment in Great Britain, Dickens et al. (1999) found that minimum wages

significantly compress the distribution of earnings, but do not have a negative

impact on employment. Moreover, in a study of the European labor market,

Machin & Manning (1997) found little evidence that minimum wages have a `bad’

effect on jobs and some evidence that they have an equalizing impact on the

distribution of income among families.

Raising the minimum wage may have an effect on wage inequality and poverty.

If earners of the minimum wage are distributed uniformly across the income

distribution, increasing the minimum wage may not affect family income inequality

or poverty rates. Even & Macpherson (1996), and Sloane & Theodossiou (1996)

have established that minimum wage earners are too widely dispersed across the

income distribution to have any influence on poverty rates or family income

inequality. Over the time period used for this study, (1984 ± 98), poverty for all

families in the US has declined at an average annual rate of about 1%, while the

growth in the real federal minimum wage has remained virtually unchanged at± 0.04 percent.2 Moreover, the simple correlation between poverty and the real

minimum wage was negative and significant, albeit small ( ± 0.11, see Fig. 1). The

relevant issue for this analysis, however, is not the simple time series correlation but

the relationship between family poverty rates and the minimum wage across states

and time, ceteris paribus.

One of the major problems with a number of studies that find that the

minimum wage influences poverty has to do with the `simulation’ of labor market

responses. This approach has arisen because of a lack of predictions regarding the

relationship between the minimum wage and the family income distribution. The

authors of these studies assume an employment elasticity with respect to the

minimum wage, which is calculated from time series studies. Since these elasticities

measure the influence of minimum wages on the employment of young adults and

not just those directly affected by the minimum wage increase, using these values

tends to minimize the disemployment effect. Moreover, many of the simulation

studies assume that all workers share in the disemployment outcome. In studies by

Horrigan & Mincy (1993) and Mincy (1990), the elasticities never exceeded one

(1) Ð which means that all low wage workers are assumed to gain from the simulated

minimum wage increase. There are also problems that arise in these studies because

of assumptions that are made concerning minimum wage coverage of workers who

M inimum Wage Policy and Poverty in the US 67

are below the existing minimum. Other factors, such as reductions in welfare

benefits, firm non-compliance, reallocations between full and part-time work, and

any offsetting labor supply changes by family members, are also ignored. These

factors tend to offset any positive impacts that a higher minimum wage may have on

poverty rates. In addition to all the above, the interaction between the `covered’ and

`uncovered’ sectors is also typically overlooked. It is suggested by theory that

minimum wage legislation affects the labor market, not just through the minimum

wage, but also by way of the relative size of the covered and uncovered sectors

(Gramlich, 1976; Brown et al., 1982). Yet, past poverty studies have omitted this

important variable.

Card & Krueger (1995) have found that states with a greater share of workers

affected by the 1990 ± 91 federal minimum wage increase had larger reductions in

family income inequality and lower poverty rates. In a paper by Lustig & McLeod

(1996), higher minimum wages were found to be associated with lower poverty rates

for a sample of developing nations, albeit they also observed a positive correlation

between higher minimum wages and unemployment. Addison & Blackburn (1999)

find a poverty reducing effect of minimum wages among teenagers and older junior

high school dropouts. It is the conclusion of many prior studies that minimum

wages and poverty are inversely related Ð this relationship being stronger in the

analyses that incorporate the simulation approach.

3. Empirical Analysis

In this study, `pooled’ data for 48 states over a 15-year period (1984 ± 98) will be

used in order to determine the significance and direction of the relationship between

minimum wage laws and poverty rates. This period is noteworthy in ascertaining the

impact of minimum wage increases. Since the federal minimum wage law remained

unchanged at$3.35 per hour between 1981 and 1989; by 1998, 18 states enacted a

state minimum wage above the federal rate.3 This variation across states during this

period provides a unique opportunity to study the effects of minimum wages.

The dependent variable is 1984 ± 98 poverty rates by state for all persons. The

choice of explanatory variables has been addressed in previous analyses of income

distribution (Partridge et al., 1996) (see Table 1). The model is log-linear and is

specified as follows:

Fig. 1. Poverty and real minimum wages (1984 = 100)

68 L. K. Stevans & D. N. Sessions

Y it = X 9itg it + «it ,

i = 1, 2, . . ., 48

t = 1984, 1985, . . ., 1998, (1)

where Y it is poverty rates, X it9 is a 1 ´ 10 vector of explanatory variables (see Table 1),

g is the 10 ´ 1 vector of parameters (including the intercept), and «it is the

disturbance term. Given the cross-sectional, time series nature of this dataset, we

will consider two variants of Equation (1) Ð one allowing for fixed effects in the

intercept across cross-sectional units or states and years. In the other model, we will

utilize fixed effects in only the intercept over time and random slope coefficients

across the cross-sectional units or states. The fixed effects model is:

Table 1. Variable explanations

Variable Source

Dependent:

Log of Poverty Rate (PR) (All Families) Bureau of the Census

Explanatory:

M inim um Wages:

Log of Minimum Wage (MW) US Department of Labor, Monthly Labor

Review, January Issues

Log of Minimum Wage Coverage(MWC)

US Department of Labor, Minimum Wage

and M aximum Hours Under the Fair Labor

Standards Act.

Income:

Log of Real Average Wage (RWAGE) US Department of Labor, Geog raphical

Profile of Employment and Unemployment

and Wages and Employment.

Labor M arket:

Log of Labor Force Participation Rate(LFPR)

US Department of Labor, Geog raphic

Profile of Employment and Unemployment.

Log of Annual Growth in Employment(EMPLOY)

US Department of Labor, Employment

and Earnings, States and Areas.

Log of Proportion of Employment inGoods Producing Industry (GOODS)

US Department of Labor, Employment

and Earnings, States and Areas.

Dem ographic and Human Capital:

Log of Percent of Population that isFemale Headed Household (FEMALE)

Bureau of the Census, Cur rent Population

Sur veys

Log of Percent of Population that isNon-White (MINORITY)

Bureau of the Census, Cur rent Population

Sur veys

Log of Ratio of Population with FourYear College Degree to Population withHigh School Degree (RATIO)

Bureau of the Census, Cur rent Population

Sur veys

M inimum Wage Policy and Poverty in the US 69

Y it = S14

k = 1mk DYear,k + S

47

w = 1lw DState,w + X9itg + «it . (2)

DYear, k are dummy variables representing years, and DState, ware the dummy

variables representing states. The assumption here is that the differences in poverty

that exist across states and years can be captured by deterministic differences in the

constant term in the model. The state effects should control for varying labor market

opportunities across states, and the year effects should `pick-up’ any economic

recessions/expansions. The second model we will consider is one in which the slope

coefficients are allowed to vary randomly across states:

Y it = S14

k = 1mkDYear,k + X9itg i + «it .

g i = Åg + li (3)

where lt is the random state effect and Åg is the mean coefficient vector. Given that

the number of years, t = 15, is small relative to the number of states, n = 48, any

estimate of the variance of the time component would be unreliable. Therefore, it is

preferable to treat the time component as fixed.4 The first term in Equation (3)

represents the fixed year effect of the intercept, and the second term is the random

effect.

Equation (2), or the fixed effects specification, was chosen for inclusion

because it has appeared in previous studies, albeit not with the same explanatory

variables as in this current analysis.5 The second specification (Equation (3)) has

never appeared before in the minimum wage literature. Nevertheless, we find it a

rather tenuous assumption that the effects (slopes) that variables, such as the

minimum wage, minimum wage coverage, labor force participation, etc, have on

poverty rates are the same for every state in the US. Whether the coefficients should

be considered as random or fixed is an empirical question that we will return to

shortly. In any event, given the delegation of welfare benefits to the states by the US

federal government, a model specification that explicitly allows for state differences

is more appropriate for use in determining the effect that minimum wage legislation

has on poverty rates.

In Table 1, the predictor variables contained in X it9 are presented by

category: Minimum Wages, Income, Labor Market, and Demographics and

Human Capital. `MW’ is defined as the maximum of the state or federal

minimum wage rate.6 As mentioned previously, minimum wage legislation affects

the labor market not just through the minimum wage rate, but also through the

relative size of the covered and uncovered sectors (Brown et al., 1982, and

Gramlich, 1976). The coverage variable, (`MWC’ ), will differ across states due to

variations in firm size and industry type. Moreover, there are exemptions for

particular industries and small firms. Minimum wage coverage is measured by

using the proportion of a state’s non-supervisory labor force that is covered by the

federal minimum wage. It is interesting to note that the variation in coverage

among the states was greater than the variation between years. In other words,

during the period studied, there was a good deal of dispersion in coverage within

the states for any given year, but this dispersion did not change much over the

period 1984 ± 98.7 To date, no other study in the subject area of minimum wages

and poverty has used a coverage variable.

70 L. K. Stevans & D. N. Sessions

The effect that these variables, MW and MWC, have on poverty rates is the

central inquiry of this study, ceteris paribus. While one would expect that higher

minimum wages and/or greater coverage would reduce poverty, there may be

offsetting employment effects that may increase poverty rates Ð it is an empirical

question.

The relationship between poverty rates and average earnings should be an

inverse one, ceteris paribus. The log of the average real wage is thus included in our

poverty regression.

There are three variables used in the Labor Force category: labor force

participation, annual employment growth, and the proportion of employment in

goods-producing industries. Greater labor force participation should lead to lower

poverty rates. Moreover, since economic growth has been found to disproportion-

ately benefit lower income groups (Bartik, 1996), the model specification includes

the annual growth in non-farm employment (EMPLOY). Goods producing jobs

provide less-skilled workers access to higher paying jobs. Thus, the proportion of

employment in goods-producing industries should be inversely related to poverty.

There are also demographic and human capital variables included in this

analysis: the percentage of female-headed households (FEMALE), the percentage

of the population that are non-white (MINORITY), and the ratio of college to high

school graduates (RATIO). We would expect the variables FEMALE and

MINORITY to be positively related to poverty and the human capital variable

(RATIO) to be inversely related to state poverty rates. However, the empirical

results associated with these variables should be interpreted with vigilance, since all

of the demographic and human capital variables had to be computed using the

1984 ± 98 Annual Demographic Surveys, March Supplements.

4. Results

The results of the estimation of Equations (2) and (3) may be found in Table 2. The

fixed effects model (Equation (2)) was estimated using ordinary least squares

(OLS), while the random coefficients model (Equation (3)) was estimated using

generalized least squares (GLS).8 The coefficients presented are estimates of the

mean coefficient vector Åg.9 For reasons of brevity, the actual state and/or year

individual effects for each model are not included in Table 2.10 However, partial `F’

statistics are presented, which test whether the state/year effects as a group belong

in the relevant equation. Since the original model (Equation (1)) was specified as

log-linear, all of the coefficients are to be interpreted as elasticity estimates or

percentages.

In column (1) of Table 2, the equation that most closely resembles the model

utilized by Card & Krueger (1995) was estimated. This model includes regional

dummy variables, which is consistent with their approach, but the current

specification found in Table 2 includes more observations. Nonetheless, the

minimum wage variable is statistically significant and negatively related to poverty

rates. These results are consistent with Card & Krueger’s (1995) claim that

minimum wage legislation can reduce poverty. Moreover, increasing labor-force

participation also reduces poverty.

One of the concerns raised with the simple model specification in column (1)

is that no allowance or control was made for varying economic conditions (Deere et

al., 1995). The estimation results of Equations (2) and (3) are found in columns (2)

and (3). These specifications include controls for labor market and demographic

M inimum Wage Policy and Poverty in the US 71

Table 2. Estimation results: poverty rate regressions

Pudney Test:

H0 : Variable Coefficients are Randomz = 90.45

Variables andDescriptive Statistics

(1)Coefficients Card

and Krueger

(2)Coefficients

Fixed Effects

(3)Coefficients

Random Effects

Dependent Variable:

Log of Poverty RateMean = 14.6STD = 4.8

Explanatory Variables:

Log of Minimum Wage(MW) ± 1.25 (2.05)** ± 1.18 (1.65)* ± 1.20 (1.68)*

Mean = 3.42STD = 1.15

Log of Minimum WageCoverage (MWC) ± 2.10 (3.56)*** ± 2.85 (4.21)***

Mean = 8.92STD = 1.08

Log of Real AverageWage (RWAGE) ± 0.55 (1.01) ± 0.64 (1.24)

Mean = 8.28STD = 1.23

Log of Labor ForceParticipationRate (LFPR) ± 0.58 (2.10)** ± 1.48 (2.45)*** ± 1.57 (2.85)***

Mean = 65.2STD = 4.5

Log of Annual Growthin Employment(EMPLOY) ± 0.32 (1.75)* ± 0.43 (1.70)*

Mean = 2.86STD = 2.18

Log of Proportion ofEmployment in GoodsProducingIndustry (GOODS) ± 0.48 (2.10)** ± 0.52 (2.35)**

Mean = 0.28STD = 0.07

Log of Percent ofPopulation that isFemale HeadedHousehold (FEMALE) 0.38 (1.98)** 0.45 (2.05)**

Mean = 13.7STD = 3.5

Log of Percent ofPopulation that is Non-White (MINORITY) ± 1.05 (1.18) ± 1.10 (1.09)

Mean = 10.1STD = 8.2

72 L. K. Stevans & D. N. Sessions

Table 2. Continued

Pudney Test:

H0 : Variable Coefficients are Randomz = 90.45

Variables andDescriptive Statistics

(1)Coefficients Card

and Krueger

(2)Coefficients

Fixed Effects

(3)Coefficients

Random Effects

Log of Ratio ofPopulation with FourYear College Degree toPopulation with HighSchoolDegree (RATIO) ± 0.58 (1.69)* ± 0.65 (1.77)*

Mean = 0.11STD = 0.08

West 2.13 (3.45)***

Midwest 2.01 (4.59)***

South 3.16 (5.87)***

Partial F Test ± StateFixed Effects 1.45**

Partial F Test ± YearFixed Effects 1.58* 1.61*

R-squared = 0.48 R-squared = 0.54 R-squared = 0.78

* Significant at 0.10 level** Significant at 0.05 level

*** Significant at 0.01 levelNote: Descriptive statistics are in the units of measurement of the raw data.

influences on poverty rates, along with a minimum wage coverage variable and year/

state fixed and/or random effects. Before discussing the sign and magnitude of the

individual coefficients, it is important to note the stated test result of the null

hypothesis found at the top of Table 2. The Pudney test, (Pudney, 1978), is a test of

the null hypothesis that the variable coefficients are random. The statistic `z’ is

distributed as a chi-squared variable with 81 degrees of freedom.11 It may be noted

that the null hypothesis is accepted (Table 2), implying that the random coefficients

model is more appropriate for modeling poverty rates across states than the fixed

effects model.

The minimum wage variable is inversely related to pover ty and statistically

significant Ð the coefficients being slightly smaller in our models, (columns (2) and

(3)). In addition, the degree of significance increases from 0.05 in the Card &

Krueger estimates (column (1)) to the 0.10 level in our models. The minimum wage

coverage variable in the current study is negative and strongly significant (a = 0.01),

and the relative magnitude of the coverage variable elasticity is large relative to the

minimum wage elasticity. For instance, in the random coefficients model (column

3), a 10% increase in minimum wage coverage is associated with a 29% decline in

poverty rates, while an equivalent change in the minimum wage is allied with a 12%

M inimum Wage Policy and Poverty in the US 73

decline in family poverty. In the fixed effects model, (column (2)), a 10% increase

in minimum wage coverage is associated with a 21% change in poverty, while the

same percentage increase in the minimum wage is also coupled with a 12% percent

decline in poverty. According to the above results, expanding the size of the covered

sector relative to the uncovered sector appears to be the best way to lift lower income

households out of poverty.

There are other notable results in Table 2. A 10% increase in labor force

participation is found to reduce poverty rates by about 15% in the fixed effects

model and 16% in the random coefficients model. These coefficients are also higher

than the minimum wage elasticity, which would lend support to the implementation

of policies that increase labor force participation. These would consist of short-term

policies such as the earned income tax credit, subsidizing childcare and long-term

policies that would involve the subsidization of human capital.

It was also found that, to a certain extent, states experiencing increased

economic growth have lower poverty rates, ceteris paribus. While the result of varying

economic growth is no surprise, what is prominent is the relatively small elasticity

estimates for this variable Ð a 10% increase in employment growth is associated

with a 3% decline in poverty in the fixed effects model and a 4% decline in poverty

in the random coefficients model. It is also interesting to note that states with a high

proportion of their workforce in manufacturing, experience lower poverty rates,

ceteris paribus. This inverse and significant relationship between employment in

goods-producing industries and poverty is not surprising and has been empirically

examined in the economic development literature (List & Gallet, 1999).

As mentioned previously, the individual state/year effects were not included in

Table 2. However, the statistical significance of the partial `F’ statistics that appear

at the bottom of Table 1I, indicate that there is good reason to include these fixed

effects in both models. The significance of the state effects may indicate that states

do indeed differ insofar as their labor markets and unmeasured demographic

differences are concerned. The significance of the year effects show that there are

economic factors and demographic trends affecting poverty rates for all states.

Finally, and as expected, we have found that states with a lower proportion of

female-headed households and a relatively higher educated workforce, experience

lower poverty rates, ceteris paribus. In the random coefficients model, a 10%

decrease in the proportion of female-headed households is associated with about a

5% decline in poverty rates, while a 10% increase in the ratio of college to high-

school educated workers is allied with approximately a 7% decline in poverty. The

MINORITY variable, or the percentage of population that is non-white, is found to

have no influence on poverty rates, ceteris paribus.

5. Conclusions

This study has examined the effect that minimum wage legislation has had on

poverty rates by state in the US. We have employed a methodology that employs year

and state fixed effects, random coefficients’ effects, and includes a minimum wage

coverage variable. The use of the random coefficients model and the minimum wage

coverage variable has never been considered in previous studies.

One implication of this analysis is the empirical observation that fixed effects

models that explain poverty across states are inferior to those that allow for random

coefficients.12 Not only is the null hypothesis of randomness accepted in Table 1I,

but the R-Squared in the random coefficients model (column (3)) is 1.4 times

74 L. K. Stevans & D. N. Sessions

greater than the R-Squared of the fixed effects regression model (column (2)).

Moreover, it is also interesting to note that every estimate of the mean coefficients

in the random model is greater in absolute value than its counterpart in the fixed

effects model. The implication is that the unmeasured factors affecting each state’s

labor market, as well as demographic differences, are not deterministic but random

events.

Another intriguing result of this study is that, while raising the minimum wage

does reduce poverty, it is not as effective as other policies. Expanding minimum

wage coverage and increasing labor force participation were both found to have

larger effects on poverty rates as compared to equivalent changes in the level of the

minimum wage. In testimony before the US House Education and Workforce

Committee, Jared Bernstein (an economist from the US Economic Policy Institute)

stated,

Finally, from the perspective of the working poor, the minimum wage is a

useful anti-poverty tool. It cannot and should not, however, be viewed as

a sole solution against poverty. This is primarily due to the fact that many

employable poor persons have only marginal attachments to the labor

market. As the literature evaluating minimum wages and poverty reveals

(as well as some new findings I will present regarding the last increase),

raising the minimum wage is associated with small decreases in the poverty

rate, but the poor need other income supports, such as the EITC and food

stamps.13

The empirical results in this study provide justification for this multifactor

approach toward poverty Ð expanding minimum wage coverage, encouraging

increased labor force participation, subsidizing higher education, as well as

increasing the minimum wage, should all be viewed as effective anti-poverty

programs.

Notes

This research was supported by a Summer Research Grant from the Zarb School of Business, Hofstra

University

1. These include studies by Horrigan & Mincy (1993), Mincy (1990), and Johnson & Browning

(1983).

2. The nominal minimum wage was deflated by the CPI-U, 198 2 ± 1984 = 100.

3. Alaska, California, Connecticut, Delaware, Washington DC, Hawaii, Iowa, Massachusetts, Maine,

Minnesota, New Hampshire, New Jersey, Oregon, Pennsylvania, Rhode Island, Vermont,

Washington, Wisconsin.

4. See Judge et al. (1985 ), p. 546.

5. Addison & Blackburn (1999) estimate a fixed effects model and find indications of a poverty

reducing effect of minimum wages among teenagers and older junior high school dropouts.

However, they use different explanatory variables in their reduced form regression.

6. This is the `prevailing’ minimum wage as noted by Card et al. (1994).

7. Results will be made available upon request from the authors.

8. Equation (3) is more commonly known as the Swamy random coefficient model. See Judge et al.

(1985), pp. 539 ± 545.

9. Of course, a different estimate can be produced for every variable by state. These results will be made

available upon request from the authors.

10. Results will be made available upon request from the authors.

11. z ~ x2 (k2 ), where k is the number of parameters in the model.

12. See Addison & Blackburn (1999).

13. http://www.epinet.org/webfeatures/viewpoints/minwagetestimony.html

M inimum Wage Policy and Poverty in the US 75

References

Addison, J. T. & Blackburn, M. L. (1999) Minimum wages and poverty, Industr ial and Labor Relations

Review, 52(3), pp. 393 ± 409.

Bartik, T. J. (1996) The distributional effects of local labor market demand and industrial mix: estimates

using individual panel data, Journal of Urban Economics, 40, pp. 150 ± 178.

Brown, C., Gilroy, C. & Kohen, A. (1982) The effect of the minimum wage on employment and

unemployment, Journal of Economic Literature, 20, pp. 487± 528.

Card, D. (1992) Using regional variation in wages to measure the effects of the federal minimum wage,

Industr ial and Labor Relations Review, 46, pp. 22 ± 37.

Card, D. & Krueger, A. B. (1995) Myth and Measurement: The New Economics of the Minimum Wage

(Princeton, NJ, Princeton University Press).

Card, D., Katz, L. F. & Krueger, A. B. (1994 ) Comment on David Neumark and William Wascher,

`Employment effects of minimum and subminimum wages: panel data on state minimum wage laws’ ,

Industr ial and Labor Relations Review, 47(3), pp. 487 ± 496.

Deere, D., Murphy, K. M. & Welch, F. (1995) Employment and the 1990 ± 1991 minimum wage hike,

American Economic Review, 85, pp. 232 ± 237.

Dickens, R., Machin, S. & Manning, A. (1999) The effects of minimum wages on employment: theory

and evidence from Britain, Journal of Labor Economics, 17(1), pp. 1 ± 22.

Even, W. E. & Macpherson, D. A. (1996) Consequences of minimum wage indexing, Contemporary

Economic Policy, 14, pp. 67 ± 77.

Freeman, R. B. (1996) The minimum wage as a redistributive tool, Economic Journal, 106, pp.

639 ± 649.

Gramlich, E. M. (1976) Impact of minimum wages on other wages, employment, and family incomes,

Brookings Papers on Economic Activity, 2, pp. 409 ± 451.

Horrigan, M. W. & Mincy, R. B. (1993) The minimum wage and earnings and income inequality, in: S.

Danziger & P. Gottschalk (eds), Uneven Tides: Rising Inequality in America, pp. 251± 275 (New York,

Russell Sage Foundation).

Johnson, W. R. & Browning, E. K. (1983 ) The distributional and efficiency effects of increasing the

minimum wage: a simulation, American Economic Review, 73(1), pp. 204 ± 211.

Judge, G. G., Griffiths, W. E., Carter Hill, R., Lutkepohl, H. & Lee, T. C. (1985 ) The Theory and Practice

of Econometr ics, (New York, Wiley).

List, J. A. & Gallet, C. A. (1999 ) The Kuznets curve: what happens after the inverted-U, Review of

Development Economics, 3(12), pp. 200 ± 206.

Lustig, N. & McLeod, D. (1996) Minimum wages and poverty in developing countries, http:ne-

tec.mcc.ac.uk/WoPEc/data/Papers/wopbriedp125.html.

Machin, S. & Manning, A. (1997 ) Minimum wages and economic outcomes in Europe, European

Economic Review, 41(3 ± 5), pp. 733 ± 742.

Mincy, R. B. (1990) Raising the minimum wage: effects on family poverty, Monthly Labor Review,

113(7), pp. 18 ± 25.

Partridge, M. D., Rickman, D. S. & Levernier, W. (1996) Trends in U.S. income inequality: evidence

from a panel of states, The Quar terly Review of Economics and Finance, 36, pp. 17± 38.

Pudney, S. E. (1978 ) The estimation and testing of some error components models, London School of

Economics, mimeo, in: G. G. Judge, W. E. Griffiths, R. Carter Hill, H. Lutkepohl & T. C. Lee (1985)

The Theor y and Practice of Econometr ics (New York, Wiley).

Sloane, P. J. & Theodossiou, I. (1996) Earnings mobility, family income, and low pay, The Economic

Journal, 106, pp. 657 ± 666.