The Tipped Minimum Wage

Agency decision

Ask Donna

What actually matters in this document.

Text

Ishaana Talesara∗

Arthur Wickard†

August 2025

Abstract

We study the tipped minimum wage and the minimum wage using US tax data.

Leveraging state policy changes, we show that the tipped minimum wage increases

base wages of tipped workers, but decreases tips, at least fully ofsetting any increase in

earnings. By contrast, the minimum wage causes earnings of tipped workers to increase

mostly through tips. Changes in tips are driven primarily by changes in the percentage

of revenue tipped rather than changes in revenue per worker or tip pooling. Changes in

tips are similar for workers throughout the frm wage distribution, suggesting frm-level

policies matter for tip rates. We fnd negative efects of the tipped minimum wage on

employment and revenue. A monopsony model where tips and wages are imperfect

substitutes to the frm can rationalize these results.

∗

Department of Economics, Massachusetts Institute of Technology. talesara@mit.edu

Department of Economics, Massachusetts Institute of Technology. wickard@mit.edu

We thank those we have worked with at the IRS, including Kelly Dauberman, Maria Orfanakis, Mark

Stanton and others. We are grateful to Daron Acemoglu, Isaiah Andrews, Josh Angrist, David Autor, Amy

Finkelstein, Nathan Hendren, Simon Jäger, Nathan Lazarus, Dean Li, Ashesh Rambachan, Nina Roussille,

Anna Salomons, Ian Sapollnik, Bradley Setzler, Anna Stansbury, Iván Werning, Christian Wolf, and participants at the MIT Labor and Macro Lunches for comments at various stages of the project. We are very

grateful to Ursina Schaede for pointing us to the call for proposals. This material is based upon work supported by the National Science Foundation Graduate Research Fellowship under Grant No. 1745302. Any

opinion, fndings, and conclusions or recommendations expressed in this material are those of the authors(s)

and do not necessarily refect the views of the National Science Foundation. The fndings, interpretations

and conclusions expressed in this paper are entirely those of the authors and do not necessarily refect the

views or the ofcial positions of the U.S. Department of the Treasury or the Internal Revenue Service. All

results have been reviewed to ensure that no confdential information is disclosed.

†

Introduction

The restaurant sector is a large part of the modern US economy and is a particularly important source of employment for low-wage workers. In 2024, restaurants and bars employed

8.5% of all private-sector workers and nearly one-ffth of workers in the bottom decile of

the earnings distribution. Unlike most low-wage industries, restaurants fnance a substantial

share of labor costs through customer tips rather than wages. Understanding how this compensation structure interacts with wage-foor regulations is therefore central to the design of

efective labor market policy.

In this paper, we use rich tax data to provide new insights into the efects of the tipped

minimum wage and the minimum wage in the restaurant sector by leveraging state-level

variation in the policies. Tip pay in the United States is governed by a two-tier wage foor.

Employers may pay tipped employees a sub-minimum base wage so long as the sum of base

pay and reported tips equals or exceeds the statutory minimum wage. Tipped worker pay is

thus jointly constrained by the tipped minimum wage (TMW) and the standard minimum

wage (MW). When the tipped minimum wage rises, restaurants must increase base wages for

workers whose base pay would otherwise fall below the new threshold. However, if they can

adjust tips, they can ofset some or all of that cost by reducing tip income. Firms’ ability

to decrease tips will critically determine the efect of the policy. Whether higher tipped

minimum wages raise total compensation or merely reallocate income between base wages

and tips is therefore an empirical question.

Policy debate over the tipped minimum wage is ongoing and contentious. In Washington

DC, Initiative 82, approved by nearly 74% of voters in November 2022, planned to eliminate

the tip credit and raise the tipped minimum wage up to the standard wage by 2027. However,

the policy’s rollout caused pushback: the DC Council recently voted to slow down and

ultimately cap the increases. Meanwhile, Massachusetts voters rejected a similar ballot

measure in November 2024. Michigan has seen ongoing battles over the tipped minimum

wage since 2018, with voters, the legislature, and the courts repeatedly overturning each

other.

Proponents of increasing the tipped minimum wage argue that tip pay is not the same as

base pay. Tips leave tipped workers vulnerable to harassment and wage theft (Schweitzer,

2021). It may also be undesirable for customers who dislike the pressure and opaque costs

induced by American “tipping culture”. Tips may increase earnings volatility, exposing

workers to more risk. Additionally, some argue that raising the tipped minimum wage

will increase earnings. One Fair Wage, a national advocacy organization campaigning to

eliminate the tip credit, uses the tagline “full minimum wage with tips on top.”1

Opponents counter that restaurants may respond to higher base wages for tipped workers

by adding services fees which are not tips,2 pooling tips, raising prices, and cutting jobs. If

restaurants are able to reduce tips to counteract the increase in base wages (by encouraging

lower tips directly, adding service fees or sharing tips among more workers), earnings could

fall, especially for those whose income is primarily tips. If they are unable to change tips, the

tipped minimum wage may function like a minimum wage, albeit one that can afect workers

with earnings above the standard minimum wage. Some restaurants claim the size of many

proposed tipped minimum wage changes will increase their wage bills by an unsustainable

amount, potentially causing them to close locations.

This paper brings new evidence and theory to the debate. A key question underlying the

debate is whether frms can adjust tips in response to the policy. Some argue they cannot—in

which case increasing the tipped minimum wage should increase earnings while decreasing the

share of earnings that is in tips. The tipped minimum wage may then function similar to the

minimum wage (while afecting diferent workers). We fnd the opposite: tips decline sharply

by 4% in response to a 10% increase in the tipped minimum wage. Base wages of tipped

workers increase (consistent with a binding policy change), so the net efect on reported

earnings (tips and base wages) is null. If tips are under-reported relative to base wages, the

efect on earnings is negative. Consistent with the view that tips can adjust in response to

policies, we also fnd that the minimum wage increases the earnings of tipped workers entirely

through tips. The two policies also have diferent distributional consequences. The tipped

minimum wage tends to increase base wages for workers in the middle of the frm earnings

distribution, whereas the minimum wage increases base wages for those at the bottom of the

distribution.

In other words, we fnd that tips and, therefore, compensation structure respond strongly

to the tipped minimum wage and the minimum wage. This produces the counterintuitive

result that earnings can in fact decrease in response to the tipped minimum wage. This

suggests that supporting the tipped minimum wage because it increases the earnings of

tipped workers may be misguided. We also fnd that tips move similarly for workers across

a frm regardless of their position in the frm earnings distribution. This suggests tips are

changed at the frm level rather than at the individual level. How do frms change tips?

Tips can change if the share of revenue that is tipped changes (the tip rate), if revenue per

worker changes, or if the number of workers who split tips changes. We fnd that changes in

Though advocates sometimes “double dip”—promising workers increased earnings while promising customers the end of mandatory tipping.

Service fees generally do not go directly to workers. Some states require that service charges do go to

workers, but allow administrative or house fees that designate that they will not go to workers.

average tips per worker are largely explained by changes in tips relative to revenue. This is

consistent with restaurants implementing service fees, changing recommended tip rates, or

shifting toward lower-tip output such as takeout.

Next, we turn to estimating employment results. It is not ex ante obvious what efect we

should expect the tipped minimum wage to have on employment. Consistent with the fact

that frms often pay both tips and base wages, tips and wages may not be perfect substitutes

for restaurants. Tips and wages may be diferent in several ways. Tip compensation provides

a form of performance pay. Tip pay may make high-revenue shifts (such as weekends and

evenings), which are otherwise undesirable due to their timing and busyness, more desirable

to workers (like busy baseball games as in Oettinger (1999)). Tips allow frms to share risk

with workers. Finally, customers may be more sensitive to higher menu prices than to paying

extra in tips (similar to tax-inclusive pricing in Chetty et al. (2009)). The strength of these

forces may vary with the level of tips.

We fnd that the tipped minimum wage has a negative efect on employment while the

minimum wage has a positive efect on employment. In line with our earnings and employment results, we fnd a positive efect of the minimum wage on total labor compensation

at the frm and a (noisy) negative efect of the tipped minimum wage on total labor compensation. Adjusting for misreporting would likely imply an even larger gap between the

two.

To rationalize our empirical fndings, we develop a simple model in which monopsonistic

frms choose both tips and wages. The efect of the tipped minimum wage depends critically

on frms’ ability to adjust tips. If tips cannot adjust, the tipped minimum wage works similar

to the minimum wage. If tips can adjust, tips fall in response to a binding tipped minimum

wage, ofsetting at least some of the increase in wages. Near the unregulated base wage,

tips do not fully ofset base wages and total earnings and employment both rise. But for a

sufciently large increase in the tipped minimum wage, earnings and employment can both

fall, which is consistent with our empirical results. Here, the mechanism is a movement

along the labor supply curve. Firms fnd it optimal to decrease tips enough such that

total earnings decreases and workers supply less labor. This stands in sharp contrast to the

textbook monopsony model, where the minimum wage can only cause declines in employment

accompanied by increases in earnings. We also show that given a binding tipped minimum

wage, frms will raise earnings entirely through tips in response to an increase in the minimum

wage.

We fnd strong evidence that tips do respond to the policies and compensation structure

afects total compensation, and therefore employment. The tipped minimum wage likely has

a negative efect on earnings and on employment. Importantly, workers may also have pref4

erences over compensation structure. Through the lens of our model, where disemployment

comes through labor supply, the empirical results imply that—even if the worker prefers

wages to tips—the overall package of lower tips, higher wages, and lower earnings makes the

marginal worker worse of.

We contribute to the literature studying the tipped minimum wage by providing the frst

frm-level evidence of the efects of the tipped minimum wage. A few papers have studied

the efects of the tipped minimum wage, generally fnding positive efects on earnings and

negative efects on employment (Wessels, 1993, 1997; Even and Macpherson, 2014; Neumark

and Yen, 2023). Allegretto and Nadler (2015) argue that the negative employment efects

are sensitive and fnd null efects on employment in their preferred specifcation. The closest

paper to ours is Jones (2016) which uses tax data to study the efect of the tipped minimum

wage on servers. This paper also fnds no efect of the tipped minimum wage on reported

earnings. The diferences across papers may be due to higher rates of tip reporting in tax

data or diferences in design. These papers generally use what Dube and Lindner (2024)

call “TWFE-log(MW)” designs, where the outcome is regressed on log(Minimum Wage) and

log(Tipped Minimum Wage) with time and state (or city) fxed efects. To our knowledge,

this is the frst paper to study the tipped minimum wage using an event study design,

allowing us to provide new estimates of the efect on earnings, wages, tips, and employment

and discuss the dynamics of these efects. In some cases, these estimates difer from the

prior literature. We are additionally able to study several new frm-level outcomes: income

dispersion within the frm, tip pooling within the frm, the tip rate, frm wage bill, and

revenue. These outcomes are important because many of the policy arguments depend

critically on whether restaurants are able to adjust tips, how they adjust tips, whether

tips can adjust independently for diferent types of workers, and how the policies afect

the dispersion of earnings across workers. Our results speak to these policy questions. An

additional contribution of ours is to discuss how the underreporting of tips afects estimates

of the earnings efect. Across all studies of the tipped minimum wage, estimates of earnings

efects are vulnerable to the underreporting of tips. By separately estimating efects on tips

and base wages, we are able to calculate the implied efects of the policy on earnings under

diferent possible reporting levels.

We also contribute to the literature studying the minimum wage by estimating the efect

of the minimum wage on base wages and tips, using a design that controls for the tipped

minimum wage, and estimating efects on wage dispersion within the frm. Restaurants have

served as the canonical laboratory for studying the minimum wage, dating back to Card

and Krueger (1994). Their prevalence in low-wage labor markets and consistent reliance on

hourly workers have made restaurant-focused studies foundational to our understanding of

the minimum wage. Yet much of this literature has had little to say about tips, despite their

central role in compensation for restaurant workers. When the minimum wage increases,

frms can comply by raising the base wages or the tips of tipped workers, which can change

the composition of earnings. The minimum wage can also afect the composition of earnings

of workers not bound by the tipped minimum wage. Given the concern of many policy

advocates about the prevalence of tip pay, policymakers may care about not only the efect

of the minimum wage on earnings, but on the composition of earnings. Knowing the efect

of the minimum wage on tips is also useful because many studies use measures of hourly

earnings that include tips. If tips are underreported relative to base wages then the estimated

change in earnings will be too low. We fnd a substantial increase in tips in response to the

tipped minimum wage, suggesting that OWEs from the minimum wage be closer to zero than

previously estimated in the restaurant industry. Additionally, estimates of the minimum

wage on earnings, employment and revenue are useful to compare to the existing literature

as we use a diferent design that allows us to control for the dynamic efects of the tipped

minimum wage. Since the minimum wage and tipped minimum wage often move together,

this can generate biases, potentially changing results substantially when the two policies

move an outcome in opposite directions. Finally, our frm level data allow us to generate

new estimates of the efect of the minimum wage on within frm wage dispersion and discuss

how diferences between tipped workers and non-tipped workers may explain the efects.

The rest of the paper proceeds as follows. Section 2 describes the setting and policy

environment we study. Section 3 describes the data. Section 4 describes the empirical

strategy. Sections 5, 6, 7, and 8 present the empirical results. Section 9 presents the model.

Section 10 discusses. Section 11 concludes.

Setting

In 2024, restaurants and bars accounted for 8.5% of all private employment in the US. This

employment is particularly concentrated near the bottom of the wage distribution. Figure 1

plots the share of workers employed in restaurants and bars for each decile of the (state level)

income distribution using the 2023 ACS. In the bottom decile of the income distribution,

18% of workers’ main industry is restaurants or bars. In the second decile, 13% work in

restaurants or bars. Restaurant work is also most prevalent at the bottom of the education

distribution, with 18% workers with less than a high school diploma and 7.5% of workers

with a high school diploma without a bachelor’s degree working in restaurants/bars.

Given this, it is no surprise that policymakers have focused on regulation intended to

help workers in this industry. In addition to the prevalence of low wage workers in the

restaurant industry, restaurants difer from most other sectors in the U.S. economy because

a substantial portion of workers’ compensation comes directly from customer tips. At the

average full-service restaurant in our data, in 2019, 45% of W-2s received any tips, and

among these workers, on average, 50% of their income was in tips. At bars, 48% of W-2s

received any tips, and tipped workers received 48% of their income in tips. Even in limited

service restaurants, 22% of W-2s received any tips, and tipped workers still received 30% of

their income in tips (see Table 1). These are likely lower bounds on the prevalence of tip

compensation given underreporting in tips relative to base wages. This reliance on tip-based

pay raises distinct concerns, such as increased vulnerability to harassment from customers,

unpredictability in earnings, and potential difculties enforcing wage laws. These issues

underscore the importance of carefully understanding the costs and benefts of a tippingbased pay system and evaluating the policies that regulate it.

How is the pay of tipped employees regulated in the United States? In most states,

employers of tipped workers are allowed to claim a tip credit and pay their employees a base

wage which is beneath the minimum wage. The diference between the minimum wage and

the maximum tip credit is the tipped minimum wage: the foor on base wage compensation

for tipped employees.

Tipped workers are still covered by the minimum wage. The total of any employee’s

tip and base wage pay must average above the minimum wage across a given time period

(often a week). If it falls short, the employer is legally required to make up the diference.

For example, the federal tipped minimum wage in 2019 was $2.13 an hour and the federal

minimum wage was $7.25 an hour. This would allow a restaurant to pay its waiters a $2.13

base wage so long as the waiters average over $5.12 an hour in tips each workweek. Notably,

this means that individuals can be bound by the tipped minimum wage but make more than

the minimum wage.

Increasing the minimum wage has an unambiguously positive efect on the hourly compensation of workers bound at the minimum wage. But raising the tipped minimum wage

does not mechanically increase the pay of tipped workers. A higher tipped minimum wage

would require some frms to increase the base pay of their tipped workers, but changes to

tip pay could counteract—or fully fip—this boost to earnings.

In this paper, we leverage variation across states and over time in tipped minimum wages.

The tipped minimum wage has its roots in the 1966 amendments to the Fair Labor Standards

Acts, which introduced the idea of tip credit and formally allowed employers to pay tipped

workers a lower base wage on the condition that tips would make up the diference to the

regular minimum wage. The federal tipped minimum wage has remained at $2.13 per hour

since 1991. While some states follow the federal standard, others have enacted higher tipped

minimum wages or eliminated the tip credit altogether. In 2019, out of 50 states and DC,

18 had tipped minimum wages at the federal level, 27 had tipped minimum wages between

the federal level and their state minimum wage, and 6 had tipped minimum wages equal to

their minimum wage. Figure 2 shows the ratio of the tipped minimum wage to the minimum

wage in 2019.

Data

3.1

IRS tax data

Our primary data source comes from the US Internal Revenue Service. Our data on worker

wages and tips comes from W-2s and our data on frm revenue comes from forms 1120S, 1065, Schedule C, and 1120. We have two samples. Our primary sample is a sample of

restaurants. The restaurant sample contains all frms that report NAICS 722511 (full service

restaurants) or 722512 (limited service restaurants) at any point from 2012 to 2019 and all

frms that report 722410 (bars) at any point from 2003 to 2019. For those frms, we have

the frm tax forms and W-2s of all workers who report at least $5,000 in annual earnings

associated with that frm for all years from 2003 to 2019. We restrict our analysis to frms

with at least three workers in a year. This restaurant sample is our primary analysis sample.

Our second sample is a frm 5% sample, where a 5% sample of all frms between 2000 and

2019 is drawn. For any frm drawn, we have the frm tax forms for all years from 2000 to 2019

and the W-2s for all workers who report at least $5,000 in annual earnings associated with

that frm from 2000 to 2019. We use this sample for placebo tests. While one may worry

about underreporting of tips to the IRS, Basker et al. (2024) show that tip reporting rates

are several times higher in tax data than in survey data such as SIPP. We discuss throughout

the paper how the underreporting of tips afects the interpretation of our estimates.

There are several desirable features of this data. First, it measures tips rather than just

total earnings or just hourly base wages. This allows us to decompose earnings efects into

tips and base wages and adjust for the possible underreporting of tips. Second, we have a

frm panel linked to individual data. This allows us to study the efects of the policies at

the frm level and to study efects across diferent types of workers (tipped and non-tipped,

high earners and low earners, etc.).

Our wage and tips measures come from W-2s. Social security tips (box 7) include all

reported tips as long as the total social security wages and tips are below the social security

cap.3 Earnings come from box 1. We defne base wages as earnings minus tips.4 Note that,

as in all studies using tax data, we observe annual earnings not hourly wages. This allows

us to measure some important objects of interest, such as wagebill and share of income that

is tips, well but makes it difcult to study hourly earnings. We restrict our earnings, base

wages, and tips results to a sample of “full-year” workers (workers who were at the frm

in the previous and the subsequent year, who are therefore unlikely to have been hired or

separated in that year). This allows us to focus on earnings efects which are not due to

changes in churn induced by the policies. However, we also report the efects on the sample

of all workers in section 8.

We construct several variables. We construct earnings and base wages for tipped workers.

Since we observe all W-2s for frms within our sample, we are able to construct the share of

workers making tips as the number of W-2s with any reported tips over the total number of

W-2s. We construct the share of earnings in tips as tips over total earnings. We defne the

compensation bill as total earnings summed across all W-2s at the frm(including tips), the

wage bill as base wages summed up across W-2s at the frm, and the tip bill as tips summed

up across W-2s at the frm. To study dispersion of earnings across workers within the frm,

we generate wages, tips and earnings by quartile of frm-state-year earnings (for all frms

with at least 4 full-year W-2s) as in Kline et al. (2019).

Our measure of revenue is the “gross receipts” line item on forms 1120-S, 1065, Schedule

C, and 1120.5 We also construct the ratio of total frm tips to frm revenue.6 Table 1 provides

summary statistics of frm-level variables for all bars, full-service restaurants, and limitedservice restaurants in 2019. Tables 2 through 5 provide frm-state-level summary statistics

for each regression sample and all outcomes.

3.2

State policy variation in TMW and MW

We construct our dataset of state tipped minimum wage and minimum wage laws using the

historical tables on “Minimum Wages for Tipped Employees” available through the U.S.

Department of Labor. Industry-specifc regulations are especially common for the tipped

minimum wage, with many states adjusting the allowed tip credit based on the tipping

norms within a sector. We drop data on regulations for industries outside of our sample.

For example, we remove the (tipped) minimum wage for “Chambermaids” in New York or

In 2025, the social security cap was $176,100. This cap is quite high for the group of workers we are

studying.

We restrict to workers with non-negative earnings, base wages, and tips and trim the top and bottom

1% for all measures

We restrict to frms with non-negative revenue and then trim the top and bottom 1%

We additionally trim the top and bottom 1% of this value.

“All Other Employees” in Vermont (which specifes a diferent policy for restaurant workers

and those who frequently receive tips).

States may also set tipped minimum wage or minimum wage levels which vary by other

frm attributes—e.g. higher for larger frms or lower for frms which ofer health insurance.

We standardize these frm types across years within a state.

Due to these targeted regulations, our fnal dataset may contain multiple diferent levels

of the (tipped) minimum wage within a state-year that could reasonably apply to frms in

our sample of full service restaurants, limited service restaurants, or bars. We consider a

state to have had a (tipped) minimum wage policy change if the (tipped) minimum wage

was increased in any of these categories within that year.

Empirical strategy

We leverage state-level variation in the tipped minimum wage and minimum wage in an

event study design similar to Cengiz et al. (2019). We consider events from 2003 to 2019

where the state (tipped) minimum wage increased by at least 5%. During this time period,

the federal minimum wage was increased which induced a higher minimum wage in several

states whose state minimum wage was equal to the federal minimum wage. In our primary

specifcation, we consider these to be minimum wage events, but our results are robust to

excluding these events caused by federal increase. Appendix Figure A1 shows the events we

use in our main specifcation.

We estimate the following regression equation to study the efects of (tipped) minimum

wage events on frm-state outcomes Yist :

Yist =

k=−3

T MW

βkT M W Ist,k

βkM W Ist,k

+ γis + δt + Ωst + εist

(1)

k=−3

T MW

where Ist,k

is an indicator which is 1 when state s had a tipped minimum wage event

k years relative to time t and 0 otherwise and Ist,k

is defned similar for the minimum wage

events. γis is a fxed efect for frm-state is and δt is a time fxed efect. Ωst denotes controls

for small changes to the (tipped) minimum wage.7

By including event time dummies for both the minimum wage and the tipped minimum

wage, our event study design can be thought of estimating the efect of the minimum wage

Our events exclude small changes to the (tipped) minimum wage. We control for two sets of dummies of

EARLY, PRE, and POST for small changes in the minimum wage and small changes in the tipped minimum

wage. Each is defned as in Cengiz et al. (2019). EARLY = 1 if −3 ≤ k ≤ −2, PRE = 1 if k = −1, and

POST = 1 if 0 ≤ k ≤ 4

events controlling for the dynamic efects of changes to the tipped minimum wage—or vice

versa. This is important because the two policies sometimes move together. This means

we are studying the minimum wage using a diferent design than previous minimum wage

papers, because we control for dynamics in the tipped minimum wage. We weight these

regressions by frm-state employment and cluster standard errors at the state level. As in

T MW

Cengiz et al. (2019), we are interested in βkM W − β−1

and βkT M W − β−1

for −3 ≤ k ≤ 4

in our event window. Our event study fgures plot these diferences.

Our main earnings and employment outcomes can all be constructed at the frm-state

level. When studying outcomes that are frm-level (revenue, tip rate), we restrict to frms

that had at least 80% of their W-2s in a single state in every year in our sample. Tables 2 and

3 compare frms in these samples to each other. As expected, single state frms are smaller,

but they are very similar to the overall sample in terms of the share of workers tipped (about

60%), the share of income in tips (48%), and the ratio of tips to revenue (0.09). Though

these frms may be diferent in unobservable ways, we fnd that our main results replicate in

this sample.

To elucidate our empirical strategy, we estimate a state level version of Equation 1 where

the outcome is log(minimum wage) or log(tipped minimum wage).

log((Tipped) Minimum Wage)st =

T MW

βkT M W Ist,k

k=−3

βkM W Ist,k

+γs +δt +Ωst +εst (2)

k=−3

We weight by state employment in “Food Services and Drinking Places” (from Statistics of

U.S. Businesses) and cluster standard errors at the state level. Figure 3 plots βkM W − β−1

T MW

and βkT M W − β−1

for −3 ≤ k ≤ 4. Figure 3a shows the efect of the tipped minimum wage

events on the minimum wage and the tipped minimum wage. As desired, the events increase

the tipped minimum wage by around 10% and have no detectable efect on the minimum

wage. Figure 3b shows the efect of the minimum wage events on the minimum wage and

the tipped minimum wage. Analogously, these events increase the minimum wage by around

10% but not the tipped minimum wage. These plots are also useful to benchmark the efect

sizes we fnd and construct elasticities, for example, of earnings with respect to the minimum

wage.

Sometimes policy changes are phased in over the course of several years. In our main

specifcation, any change above 5% is counted as a separate event. We can alternatively

collapse consecutive increases into a single event marked by the frst year of the changes. We

Note that, in minimum wage designs, β−1 is generally not collinear with other coefcients due to repeated

events.

show robustness to this alternative event specifcation in the appendix.

Earnings results

To understand the efect of the minimum wage and tipped minimum wage on the earnings

of restaurant workers, we study these policies’ impact on base wage and reported tip income

and discuss how these results can be aggregated into an overall earnings efect.

5.1

Base wages and tips

We can make the average base wages for tipped and non-tipped workers at the frm-state

level. First, we estimate the efect of the tipped minimum wage events on the base wages

of tipped workers. Figure 4 shows that a 10% increase tipped minimum wage increases

base wages by around 3.5% throughout the post period which is consistent with the tipped

minimum wage, which regulates base wages, binding. Next, we estimate the efect of the

tipped minimum wage on tips. The tipped minimum wage does not regulate total earnings,

so if restaurants are able to reduce tips in response to an increase in the tipped minimum

wage, earnings could stay the same, or even decrease in response to the tipped minimum

wage. We fnd that reported tips decrease by about 4% in response to the tipped minimum

wage.

In contrast, the minimum wage does not have a signifcant impact on the base wages

of tipped workers, and a 10% increase in the minimum wage causes a large and persistent

increase (about 5%) to the average tips reported by tipped workers. This is consistent with

the fact that the minimum wage regulates the total compensation of tipped workers, but not

the base wages.

Figure 4 further emphasizes the diferent efects of the tipped minimum wage and minimum wage on the composition of earnings by plotting the event study for the efect of these

policies on the tip share of reported income for tipped workers and for all workers. As expected, the minimum wage caused tips to be a larger share of income whereas the tipped

minimum wage decreased the tip share. Since the share of earnings that are tips can be

calculated without the full-year restriction, this provides a useful complement to previous

results.

These results answer one of the key questions posed by the policy debate and the model:

frms can and do change tips in response to the tipped minimum wage and the minimum

wage. This raises the question of how tips adjust in response to these policies. Changes

in revenue per worker could pass through into tips, or frms could actively add service fees,

change tasks performed, or change the pool of workers sharing tips. We discuss and attempt

to disentangle diferent mechanisms in section 7.

5.2

Earnings

What do these results say about the efect of the policies on total earnings? We can sum

tip income and base wage income to estimate the efect of (tipped) minimum wage events

on reported earnings. Figure 4 also demonstrates that the tipped minimum wage had no

detectable efect on the reported earnings of tipped workers, while the minimum wage increased tipped worker reported earnings by around 1.5% in the post period. However,

especially among tipped workers, reported earnings are not necessarily the same as true

earnings. If workers underreport tips, this means that the efect on tip income is relatively

more important for determining the efect on total earnings than we are measuring. To see

this point, let true earnings equal base wages plus tips

E =B+T

(3)

Suppose that all of base wages are reported and a fraction r ∈ (0, 1) of tips are reported.

Let α be the fraction of true earnings that are tips before the policy.

Ereport = B + rT = (1 − α)E + rαE

(4)

Now, suppose the policy changes tip income by t and base wages by b such that

T ′ = (1 + t)T, B ′ = (1 + b)B

(5)

Then, the percent change in true earnings is

%∆E = b(1 − α) + tα

(6)

And the percent change in reported earnings is

%∆Ereport = b

1−α

rα

(1 − α) + rα

(7)

That is, the efect on tips is down-weighted relative to the efect on base wages. Here, we

are considering the efects of constant under-reporting and assuming r does not change in

response to the policy.9 If not all tips are reported (r < 1), then the efect of the minimum

wage on earnings is larger than its positive efect on reported earnings. Similarly, the efect of

the tipped minimum wage on earnings would be lower than its efect on reported earnings. In

fact, any amount of under-reporting tips (relative to base wages) would suggest a negative

efect of the tipped minimum wage on earnings. The average frm in full service, limited

service, and bars reports that tipped workers receive 46% percent of their total income in

tips on average. If the true percent of their income coming from tips were any larger, then the

total efect on earnings would be negative. This supports concerns among opponents of the

tipped minimum wage that the policy may decrease earnings for some workers. The idea that

tips would decline by more than wages increase is somewhat counterintuitive. It suggests

that tips and wages are imperfect substitutes to the frm as a method of compensation. We

formalize this idea and provide conditions under which it can happen in the model in section

5.3

Efects by within-frm earnings quartile

Another important question for policymakers is whose earnings are being afected. The

minimum wage targets those at the bottom of the wage distribution. However, the tipped

minimum wage targets tipped workers. Tipped workers are often the subject of policy

attention due to the concern that they are low earners. However, the tipped minimum wage

can be binding for tipped workers who earn more than the minimum wage. In fact, the

tipped minimum wage can be binding for high-earning tipped workers but not low earning

tipped workers, for example, if high earning tipped workers are servers and low-earning

tipped workers are bussers who receive a smaller share of earnings through tips through

server tip outs. We separate workers in each frm-state-year with at least four full-year W-2s

into quartiles based on annual reported earnings. This analysis is on all workers, not just

tipped workers, since an important way that the minimum wage can increase earnings is

by increasing the wages of non-tipped workers. Figure 5 shows the efects of the policies

on base wages and earnings of full-year workers in each quartile of the frm-state annual

earnings distribution. The minimum wage raises base wages primarily in the bottom quartile,

and increases reporting earnings in the bottom two quartiles. The tipped minimum wage

increases base wages in quartiles 2 and 3, suggesting that tipped workers bound by the tipped

minimum wage are on average higher up in the frm earnings distribution than minimum

wage workers. Figure 5 also shows the efects on tips in each quartile (for full-year tipped

workers). We fnd that changes in tips are remarkably similar across quartiles. This suggests

If reporting changes in response to the policy, that could have very diferent efects. The next section

discusses why we do not think changes in r are driving our results.

tips are changed via frm-wide changes and not just for workers bound by the policies. In

fact, the (noisy) point estimates suggest that the tipped minimum wage may even decrease

earnings for quartile 1. This could be explained by quartile 1 having many minimum wage

workers who make some tips (e.g. bussers). They would not be bound by the increase in

the tipped minimum wage, but they would be afected by the establishment-wide decline in

tips.

The common change in tips across quartiles also suggests that our results are not driven

by compliance. While the IRS does not enforce minimum wage compliance, it is reasonable

to ask if our results are driven by changing compliance motivations causing workers to report

fewer tips following an increase in the tipped minimum wage (i.e. is r in Equation 4 changing

in response to the policy?).10 Specifcally, if frms encouraged workers to report enough tip

income to have their hourly pay reach the minimum wage, then increased base pay due

to the tipped minimum might lead workers who were near this threshold to report less tip

income. Similarly, frms could encourage workers to report more of their tips in response to

the minimum wage. This force would be weaker for higher income workers who are more

likely to already be reporting income well above the minimum wage. However, we see that

the efect on tips is similar across the distribution of full-year earnings. This makes it unlikely

that our results are driven by changing tip reporting to comply with the policy and instead

suggests that the cause of the decline in tips is an establishment-level efect such as the

introduction of service fees or changes to average restaurant trafc.

Employment

Now, we turn to asking what the efect of the tipped minimum wage and minimum wage

is on restaurant employment. It is not obvious what efect we would expect the tipped

minimum wage to have on employment, even given the efects on earnings. It depends on

worker preferences over tips versus base wages and the frm’s relative cost of paying the two

types of compensation.

6.1

Number of W-2s

First, we look at the efects of the policies on log W-2 counts. Similar to the earnings results,

this outcome is at the frm-state level. In Figure 6, we see that in response to the minimum

wage, the number of W-2s increased by about 2%. In contrast, the number of W-2s fell

Basker et al. (2024) show a bunching pattern of tips consistent with some reporting behavior being

afected by compliance

substantially in response to the tipped minimum wage—reaching a 2.5% decline by the end

of our event window.

While the W-2 count is often used as a measure of employment (Kline, Petkova, Williams,

and Zidar, 2019; Lamadon, Mogstad, and Setzler, 2022), it measures the total number of

workers who were employed by the frm at any point in the year which is diferent from most

employment measures which count the number of employees at one snapshot in time. This

means that the W-2 count is infuenced by retention efects which can impact the number

of workers who cycle through the frm throughout the year. Indeed, we show in Appendix

Figure A2 that the minimum wage caused an increase to both hires and separations while

the tipped minimum wage caused a decrease to both hires and separations11 . This means

that the W-2 count results may be at least partially driven by a negative churn efect for

the minimum wage and a positive churn efect for the tipped minimum wage. However, the

efects on total labor payments at the frm in Section 6.2 and the efect on revenue in Section

7.1 suggest an important role for (dis)employment efects.

6.2

Wage bill

Studying the efect on total earnings across all workers at the frm also helps us understand

employment efects. This is often called the wage bill. We defne the wage bill to include

base wages only, the tip bill to include tip income only, and the compensation bill to include

both forms of compensation. The efect on the compensation bill jointly summarizes the

efect on earnings and employment, including efects on hours or retention. The minimum

wage literature often asks how much of an earnings increase is counteracted by a decrease in

employment (Dube and Lindner, 2024). If earnings increase, the compensation bill measures

this object. Wage bill, tip bill, and compensation bill are calculated using all workers at

the restaurant, without the full-year restriction, to fully capture efects driven by changes in

retention, hours, employment, and hourly wages.

Figure 7 shows the efects of the policies on the wage bill, tip bill, and compensation bill.

Since the wage bill and compensation bill include non-tipped workers, we include estimates

of our main earnings regression on both tipped and non-tipped workers in Appendix Figure

A3. Taking the point estimates, the TMW decreases the compensation bill by about 1.8%,

though the estimates are somewhat noisy. That is, the TMW lowers total payments to

workers. The fact that the compensation bill declines despite no efect on average earnings

is consistent with a disemployment channel.12 The MW increases the compensation bill by

Importantly, we cannot tell whether separations are voluntary or involuntary, so interpreting these results

as evidence of labor supply or labor demand changes is difcult.

We do not think decreased churn explains these results since we estimate efects on average earnings

almost 3% while increasing average earnings by less than a percent, consistent with positive

employment efects. Similarly, the TMW decreases the tip bill by 6% by the end of the post

period and the MW increases the tip bill by around 7%. Much like in our average earnings

results, if tips are consistently underreported relative to base wages, we will underestimate

the decline in the compensation bill, since the tip bill decreases by more than the wage bill.

The employment and earnings results are consistent with a large tipped minimum wage

increase causing frms to decrease tips by more than base wages change, decreasing total

compensation, and causing the quantity of labor supplied to the frm to fall. We formalize

this idea in section 9.

Unpacking changes in tips: revenue, tip rate, and tip

pooling

Section 5 showed that reported tips decrease in response to increases in the tipped minimum

wage and increase in response to increases in the minimum wage. These results raise the

question of how tips adjust in response to these policies. There are several possibilities. First,

frms may be able to change the average tip rate (tips as a percent of revenue). They may

do this by implementing service fees, administration fees, or kitchen fees. These fees are not

legally considered tips and are not required to be used for worker compensation.13 However,

they may decrease how much a customer tips. In Washington, DC, for example, these are

sometimes listed as “I-82 fees,” presumably to signal to consumers that the fee was caused

by the initiative to equalize the tipped minimum wage with the minimum wage. Restaurants

may also change recommended tips on receipts or electronic payment systems. Restaurants

may shift towards or away from low-tip output like takeout. Restaurants often engage in

some form of tip pooling, whereby servers share a portion of their tips with one another and

may also “tip out” to non-serving staf such as bussers, food runners, and bartenders. They

can change average tips per tipped worker by changing the number of workers in the tip

pool. Tips may also move mechanically with revenue per worker. The price efects of the

minimum wage are well-documented by the literature. If prices increases result in revenue

per worker increases, this naturally passes through into tip income. Quantity can also afect

tips through revenue per worker. If shifts become slower, revenue per worker may decrease,

decreasing tips.

We attempt to disentangle these explanations here, by estimating the efect of the policy

using full-year workers.

In a few states there is some regulation around these fees—such as requiring restaurants explicitly state

the fees will not be going to employees

tips

on revenue, tip rate ( revenue

), and tip pooling. We fnd that, for both policies, changes in

tips can mostly be explained by changes in the tip rate, suggesting that the service fee,

recommended tips, and more/less tip-intensive output explanations are more important in

explaining the changes in tips.

As described in section 4, since revenue is defned at the frm level (rather than frmstate), we restrict our analysis of revenue and tip rate to the sample of frms that had at

least 80% of their W-2s in a single state in every year in our sample. In Appendix Figure

A4, we show that our main earnings results replicate in this sample.

7.1

Revenue

We estimate the efect of the TMW and the MW on revenue at the frm. Figure 8a shows that

the tipped minimum wage decreases revenue by about 2%. Our estimates of the minimum

wage on revenue are similarly positive (point estimates of about 2%), but noisy. These

results are broadly consistent with the disemployment efects of the tipped minimum wage

and positive employment efects of the minimum wage. Consistent with the previous results,

these results underscore the idea that the TMW afects restaurants very diferently from the

MW.

The efect of the policy on revenue per worker is most relevant for explaining the efect

of the policies on average tips. Naively taking the estimates on the number of W-2s as

estimates on employment would suggest that revenue per worker increases in response to the

TMW and decreases in response to the MW, as the employment efects are larger than the

revenue efects. In this case, the efects on revenue per worker cannot explain the efects on

average tips.

However, as we have emphasized, the efects on W-2s likely include efects on churn as

well and may thus overstate the employment efects. In which case, the efect of the TMW

on revenue per worker could be negative and could contribute to lower tips. We continue

this discussion below.

7.2

Tip rate

Next, we look at the efect of the (tipped) minimum wage on the tip rate: the sum of reported

tips across all the frm’s workers divided by revenue. This is useful because it allows us to

understand whether quantity or price efects are driving our tip income results. If the

same number of workers are serving more/less customers or getting tipped the same rate on

higher/lower prices, then this will change their average tip received but will not impact the

tip rate. In other words, changes in revenue per worker would change tips without changing

the tip rate. On the other hand, if restaurants impose service fees which cause patrons to

tip less on a bill, the tip rate would decline. Figure 8b shows that the tip rate does fall in

response to the tipped minimum wage and rise in response to minimum wage and that the

magnitude of the changes is similar, but slightly smaller than the efects on average tips.

This suggests that changes in the tip rate play an important role in explaining the efect

on tips. It is worth emphasizing that seemingly small changes in the level of the tip rate

can explain these large changes in tips. If restaurants are able to change the amount tipped

on a check from 10% to 11% of revenue, all else equal, tip income (and the tip rate) would

increase by 10%.

7.3

Tip pooling

Finally, frms can change tips by changing how many workers tips are shared among. First,

we look at the share of W-2s receiving any tips. For example, frms could share the same

tips among more workers to decrease average tips per tipped worker. We fnd a small, noisy

decrease in the share of workers receiving tips in response to the tipped minimum wage and a

small, noisy increase in the share of workers receiving tips in response to the minimum wage.

These efects go in the wrong direction to explain the efect on tips through tip pooling.

There are diferent possible explanations for this result. One possible explanation is that

tipped workers are relatively cheaper as the share of their income that is tips increases. So

in response to a TMW increase, restaurants demand fewer tipped workers relative to nontipped workers. If this is happening alongside tip pooling, the efect of the policy on the

share of workers making tips may not be informative. For example, the reduction in share

of workers making tips could be evidence of larger disemployment efects for highly tipped

employees (e.g. waiters) combined with tip pooling with additional workers (bussers).

Another test for tip pooling involves looking at the efect on tips by quartile as changes

in tip pooling could show up as diferent efects on tips in diferent parts of the frm earnings

distribution. For example, we would expect paying low wage bussers a small amount of tips

from the pool to lower tips in the bottom quartile. But as shown in Figure 5 (see section

5.3), we fnd remarkably stable efects on tips across the frm earnings distribution. Taken

together, this suggests a limited role for tip pooling in explaining the tip efects of the tipped

minimum wage.

The fact we fnd limited evidence of tip pooling may be due to regulations dictating

whether employers can form a tip pool and who they can include in it. The Fair Labor

Standards Act states that if frms claim a tip credit, then they can only distribute tips

among workers who regularly receive tips but if the frm pays all workers a base wage at least

$7.25 then they can implement a “nontraditional” tip pool which includes non-managerial

back-of-house workers. But many states impose further restrictions—for example, stating

that only voluntary tip pools are allowed or expressly forbidding back-of-house employees be

included in the pool (regardless of whether the frm claims a tip credit).

7.4

Summary

Together, these results suggest that tips are mostly, but not entirely, explained by changes in

the tip rate. Changes in revenue per worker may also contribute to changes in average tips.

Tip pooling is unlikely to explain changes in tips. Combined with the previous evidence that

the tip income efect is constant across worker income quartile, this suggests that the change

in tipping behavior is largely driving by frm-level policies which impact the average tip rate

such as service fees, recommended tips, or ofering takeout versus sit down service.

Robustness

We show robustness to non-full-year workers, full service restaurants, and diferent event

defnitions as well as a placebo regression on a sample of all workers.

Our main earnings variables are all created using full-year workers, defned as those who

had a W-2 from the same frm in the previous year and in the subsequent year. This is

important because W-2s measures annual earnings which are afected by the number of days

a worker works at a frm. Changes in average time a person spends at a frm (e.g. induced

by hirings or separations) in response to the policies may change annual earnings/wages/tips

even if hourly earnings/wages/tips do not change. In fact, we fnd that hires and separations

decrease in response to the tipped minimum wage and hires and separations increase in

response to the minimum wage (see Appendix Figure A2).

Re-estimating our main earnings results on all workers produces estimates consistent with

the efects on churn. Appendix Figure A5 shows estimates including non-full-year workers.

They are broadly consistent with the main results presented in Figure 4. However, the

positive efect of the TMW on base wages (and earnings) is now larger, and the negative

efect on tips is smaller. This is consistent with the bias one would expect from the TMW

decreasing churn. Specifcally, workers spending more of the year at the frm would magnify

increases in hourly wages and mitigate decreases in hourly tips. Similarly, for the minimum

wage the positive efect on tips and earnings is now smaller and the negative efect on base

wages is larger. This is consistent with the MW increasing churn.

We estimate the main results on base wages, reported tips, and reported earnings re-

stricting to full-service restaurants. While limited-service restaurants still pay a substantial

amount in tips, it is not clear how common it is for limited-service frms to claim the tip

credit. The results are very similar. Appendix Figure A6 shows these results.

In our main specifcations, we allow binding changes to the federal minimum wage to be

counted as events (these will not afect all states, as some states will have minimum wages

above the federal minimum wage). Since there are state-level changes in the tipped minimum

wage in years where the federal minimum wage changes, including these allows us to control

for efects that may be induced by the federal minimum wage change. We show robustness

to excluding these events in Appendix Figure A7. When excluding federal changes from the

set of events, we include controls for whether a federal event happened similar to how we

control for small changes in the policies. We also show robustness to collapsing consecutive

events into a single event in Appendix Figure A8.

Finally, we estimate the efect of the minimum wage and the tipped minimum wage on

earnings in our 5% sample of all frms as a placebo test. Since minimum wage workers are

a small share of the workforce, we do not expect to be able to estimate a precise efect of

the policies on the 5% sample of frms in all industries. As expected, we fnd null efects on

earnings. Appendix fgure A9 shows these results.

Model

Next, we turn to theory to help us rationalize the empirical results. Why can the tipped

minimum wage cause earnings and employment to both fall? Under what conditions can

that occur? Why does the minimum wage raise earnings through tips? We write a simple

model where monopsonistic frms pay workers in base wages and tips. Critically, we allow

frms to adjust tips, since our empirical results suggest that frms can adjust tips. If frms

cannot adjust tips, then the tipped minimum wage operates as in the textbook minimum

wage model, strictly increasing the earnings frms must pay. In our model, we show that tips

(weakly) fall in response to increases in the tipped minimum, earnings and employment can

both fall in response to increases in the tipped minimum wage, and that in the presence of

a binding tipped minimum wage, increasing the minimum wage will increase tips and have

no efect on base wages.

The frm chooses wages and tips per worker to maximize profts π. Labor is a function

of total earnings w + T . Revenue is linear in labor, with the price that frms can set being

a function of the tips its workers receive.

π(w, T ) = P (T )L(w + T ) − wL(w + T )

(8)

This model nests many diferent ways that frms may adjust tips. For example, frms may

change the recommended tips (to increase tips). Or alternatively, frms may add a service

fee (to decrease tips), which they can claim directly as a revenue.

We assume that output is a continuously diferentiable function of tips, P ′ (T ) ≤ 0 (inducing higher tips is costly), P ′′ (T ) ≤ 0 (this cost is increasing), and P ′ (0) = 0 (around

zero there is no cost to increasing tips).14 Notably, for algebraic simplicity, we are assuming

that revenue responds to tips per worker, rather than total tips at the frm or the average

tip rate.15 We assume L′ (w + T ) > 0 and L′′ (w + T ) < 0: labor supply is upward sloping

(monopsony) and workers become harder to recruit as L increases.

P ′ (T ) ≤ 0 captures the idea that it is not costless for workers to receive tips. Customers

realize that the expectation they tip will impact their bill. Increasing T is costly because it

decreases the price P the frm can charge, lowering revenue. Increasing w is costly because

frms have to pay workers w out of revenue, lowering profts. In the unregulated equilibrium,

frms choose wages and tips to equalize the costs. They will set −P ′ (T ) = 1. Holding

constant the frm’s employment, the cost of paying a dollar of base wage pay (per worker) is

1 while the cost of a dollar of tip pay (per worker) is −P ′ (T ), the extent to which prices fall.

Since P ′ (T ) does not always equal 1, this is the sense in which wages and tips are imperfect

substitutes.

The idea that this cost is increasing (P ′′ (T ) ≤ 0) could be due to several microfoundations. First, customers’ distaste for tipping may be increasing in the level of the suggested

tip. Additionally, it may be low-cost for frms to increase tips from lower initial tipping levels

because it introduces a performance incentive. Tipped workers have an incentive to show

up for the highest demand shifts and to provide high-quality service. These benefts might

be particularly strong for the frst few dollars of tips, justifying that P ′′ (T ) ≤ 0. Lastly,

we assume that the cost of adding tips (and how that cost is changing) is 0 when there are

initially no tips. This is true if, for example, it is close to costless to allow customers who

are very eager to tip to put a few dollars in a tip jar.

The proofs of the following propositions are in Appendix A2.

Proposition 1.

wage

Tips (weakly) fall in response to increases in the (binding) tipped minimum

When a frm’s wage decision is bound by the tipped minimum wage, w̄, they solve for

You could alternatively replace 0 with a diferent lower bound for tips

Since production is linear, this is inconsequential. In another model with nonlinear production, if tips

are set as a share of revenue (rather than per worker), an additional force emerges, which is that hiring an

additional worker reduces tips per worker for all employees. Wessels (1997) models tips as a fxed share of

revenue rather than a choice and features this force.

the optimal amount of tips to ofer. This leads to the following frst order condition:

−P ′ (T )L(T + w̄) = (P (T ) − w̄)L′ (T + w̄)

(9)

Increasing tips is costly because it reduces revenue by −P ′ (T )L(T + w̄). Increasing tips

is benefcial because it increases the number of workers (where the frm profts P (T ) − w̄

on the marginal worker). Increasing w̄, the tipped minimum wage, decreases the beneft of

recruiting another worker by causing the markdowns to narrow and also reduces the number

of workers the frm hires with an added dollar of tips, L′ (T + w̄). It also increases the

marginal cost of paying tips, −P ′ (T )L(T + w̄), since L is now higher. Since the cost of

increasing tips is now higher and the beneft lower, the frm reduces tips to satisfy 9. (See

Appendix A2 for more detail).

Proposition 2. For a small increase in the tipped minimum wage above the unregulated

equilibrium wage, total earnings and employment both rise

For small increases in the tipped minimum wage (close to the unregulated base wage), the

decline in tips will not fully ofset the increase in base wages. This increase in compensation

will increase employment, as in the textbook monopsony case. Note that the decline in tips

will mute (though not reverse) the efect on earnings.

Proposition 3. If P ′′ (0) = 0, a sufciently large increase in the tipped minimum wage will

cause total earnings and employment both to fall

Recall from proposition 1, that when the tipped minimum wage increases, frms reduce

tips because the marginal cost of paying tips has gone up and the marginal beneft has gone

down. Reducing tips causes the marginal cost to go down. However, since P ′′ (T ) < 0, when

tips are low, the efect of reducing tips on the LHS of 9 is limited, so reducing the marginal

cost requires a larger reduction of T .

To see the intuition, recall that the marginal cost of tips and wages are equal in unregulated equilibrium, −P ′ (T ) = 1. The inframarginal cost of tips is always less than 1,

and is smaller the lower tips are, whereas wages always cost 1. So if the tipped minimum

wage increases, the frm reduces tips, giving up dollars of tip income that cost less than 1.

The bigger the tipped minimum wage increase, the increasingly “cheap” inframarginal tip

compensation the frm has to give up, so it requires larger declines in tips relative to the

wage increase. Wages and tips are not perfect substitutes. At low levels of tips, tips are

“cheap”—allowing generous customers to toss a buck in the tip jar as no efect on revenue—

but wages never are. (The existence of some point at which tips fall by more than wages

increase is not generically true. The condition, P ′′ (0) = 0, captures the idea that the cost of

paying tips must be falling sufciently fast near 0.)

This result is notably diferent than the frm response to the minimum wage. Here, employment has decreased not because the frm is of the labor supply curve—employment has

fallen because the tipped minimum wage reduced total compensation, reducing the quantity

supplied of labor. This case is consistent with our empirical results.

Proposition 4. If the tipped minimum wage binds, then increasing the minimum wage will

increase tips and have no efect on the base wage.

If the tipped minimum wage binds, then we have −P ′ (T ) < 1. This means that tips the

per-worker cost of paying a dollar of tips, −P ′ (T ), is strictly less than the per-worker cost of

paying a dollar of base wages, 1. When the minimum wage increases, the frm will choose to

satisfy the higher compensation requirement with an increase to tips. At the point at which

the minimum wage causes base wages to increase (in addition to tips), the tipped minimum

wage must no longer bind.

Discussion

Reconciling empirical evidence and theory To summarize the results on the tipped

minimum wage, we fnd that the tipped minimum wage increases base wages, but causes

tips to decline. We estimate a null efect on reported earnings, which suggests that if tips

are underreported total earnings likely fell. Additionally, we show that employment declines

following increases to the tipped minimum wage. What can explain a decline in earnings

and employment? Standard monopsony models of the minimum wage would predict that

employment declines follow after an increase in earnings that is large enough to push the

frm of the labor supply curve. But in the case of the tipped minimum wage, employment

declines can be driven by the frm reducing total compensation, leading the worker to reduce

labor supply. Increasing the required base wage afects the frm’s incentive to pay the workers

tips. Paying a higher base wage reduces the frm’s markdown on its marginal worker, so the

monopsonistic frm will cut tip pay, which pushes its margins back up. But as the frm

adds service fees or reduces suggested tips to decrease tip pay, it gives up an increasingly

cheap form of pay. This means that one dollar of a base wage raise will come at the cost

of an increasingly large decline to tip pay—which, for a large enough tipped minimum wage

increase, will net out to earnings falling.

What determines the beneft of reducing tips? As discussed in the model section, we

model tips as decreasing revenue to capture the fact that customer demand responds negatively to paying tips. But the particular shape of P (T ), the marginal efect of changing tips

on revenue, captures many forces. First, tips operate as a form of performance pay, which

may increase productivity. Second, frms value workers more during busy shifts (evenings,

weekends, holidays), but these shifts are the most undesirable for workers under constant

hourly pay. Tips make busy shifts desirable to workers (this could be thought of as another

form of performance pay). Third, tips provide a form of risk-sharing. If one week is slow,

workers make less money. While frms must make up pay up to the minimum wage on average, any workers making on average above the minimum wage (who may still be making

the tipped minimum wage), can experience substantial volatility in earnings. Tips might

provide a substantial source of wage fexibility in an economy with otherwise high levels of

downward nominal wage rigidity. Fourth, customers may have diferent elasticities of demand with respect to tips compared to prices. We assume the marginal beneft of reducing

tips for any of these reasons is smallest when tipping is close to zero: customers’ preferences

for paying tips may vary by the level—strongly disliking a suggested tip of 25% but hardly

noticing tipping spare change, or frms may beneft most from the frst few dollars of risk

sharing or performance pay.

It is worth noting that an important diference between the two policies that we do

not consider in the model is that the tipped minimum wage mostly afects the restaurant

sector whereas the minimum wage afects all sectors employing low wage workers. This can

generate diferent general equilibrium efects. Demand for a single sector’s products is more

price elastic than demand across many sectors. Firms may be able to increase prices with

limited efects on quantity in response to the minimum wage but not in response to the

tipped minimum wage.

Comparison to minimum wage literature The minimum wage literature often uses the

own-wage elasticity (OWE) to compare estimates across studies. Dube and Lindner (2024)

call any OWE less than -0.8 a “large negative OWE.” It is large in the sense that an elasticity

of -0.8 could be interpreted as 80% of the gains in earnings being erased by disemployment.

If employment falls due to the worker reducing labor supply as total earnings falls, as we

argue can explain the results in this paper, it is inappropriate to calculate an OWE. But we

can use our estimates for the efect on the wage bill and the tip bill to compare the size of our

efects to the literature. An OWE of -1 would correspond to 0 efect on the compensation

bill—a very large negative efect where all earnings gains are erased by employment losses.

We are unable to reject a null efect of the tipped minimum wage on the compensation

bill and the point estimate is negative. Therefore, this is a more negative efect on the

compensation bill than implied by 83% of the 70 published studies Dube and Lindner (2024)

which estimate OWEs greater than -0.8. In contrast, our estimates of the minimum wage on

earnings and employment imply a large positive OWE of over 1. This estimate is near the

top of the range presented in Dube and Lindner (2024) which may be due to the increase

in employment being partially driven by churn and the fact that, unlike prior studies, we

estimate our earnings efect controlling for changes in the tipped minimum wage (which

frequently happen concurrently with changes in the minimum wage and would otherwise

bias downwards the earnings and employment estimates).

Policy implications Workers may have preferences over compensation structure and not

just earnings. Indeed, many policy advocates are specifcally concerned that tips are an

unjust form of pay. If workers prefer base wages to tips, the wage and tip bill results would

fail to account for the fact that the compensation package improved even if earnings did

not increase in response to the tipped minimum wage. Similarly, the increase in earnings

in response to the minimum wage overstates the improvement in the compensation package.

Even if it is true that workers prefer base wages to tips, our theory suggests that the negative

employment efects of the tipped minimum wage are driven by labor supply, implying workers

do not prefer the new compensation package.16

These results inform the policy debate by providing evidence that frms are able to change

the tip rate in response to policy. Many in the policy debate assume that tips cannot be

adjusted, implying potentially larger increases in earnings for workers and higher costs for

restaurants. We fnd that tips do move in response to the policy and that they move primarily

through changes in the tip rate. Importantly, economically small changes in the level of the

tip rate can cause large changes in tip income. However, we fnd that the tip rate is a blunt

tool—tips move similarly for tipped workers in all parts of the frm’s income distribution.

Our results suggest that it is unlikely, for example, that frms just increase tips for those

making below the minimum wage when the minimum wage increases. Rather, tips increase

throughout the distribution.

Conclusion

We fnd that the tipped minimum wage reduces earnings and employment. Importantly, this

is because frms are able to change tips and choose to reduce tips by more than base wages

increase. Our model suggests that the disemployment efects are driven by labor supply—

implying that workers are worse of. This means that the tipped minimum wage is not an

Through the lens of our model, it is difcult to explain our empirical results with the interpretation that

workers prefer the new compensation package but the increased compensation causes “classic monopsony”

negative employment efects. If frms were of their labor supply curve, they could reduce tips further.

efective policy to help tipped workers. The magnitude of the welfare consequences of the

policy depends on worker preferences over compensation structure and the extent to which

other industries are able to absorb these workers and at what wages. In contrast, we fnd

that the minimum wage increases earnings in large part through tips, increasing the earnings

of tipped workers without disemployment efects.

References

Allegretto, Sylvia and Carl Nadler (2015), “Tipped wage efects on earnings and employment

in full-service restaurants.” Industrial Relations: A Journal of Economy and Society, 54,

622–647.

Basker, Emek, Lucia Foster, and Martha Stinson (2024), “Tip of the iceberg: Tip reporting

at us restaurants, 2005-2018.” Technical report.

Card, David and Alan B. Krueger (1994), “Minimum wages and employment: A case study of

the fast-food industry in new jersey and pennsylvania.” The American Economic Review,

84, 772–793.

Cengiz, Doruk, Arindrajit Dube, Attila Lindner, and Ben Zipperer (2019), “The efect of

minimum wages on low-wage jobs.” The Quarterly Journal of Economics, 134, 1405–1454.

Chetty, Raj, Adam Looney, and Kory Kroft (2009), “Salience and taxation: Theory and

evidence.” American economic review, 99, 1145–1177.

Dube, Arindrajit and Attila Lindner (2024), “Minimum wages in the 21st century.” Handbook

of Labor Economics, 5, 261–383.

Even, William E and David A Macpherson (2014), “The efect of the tipped minimum wage

on employees in the us restaurant industry.” Southern Economic Journal, 80, 633–655.

Jones, Maggie R (2016), “Measuring the efects of the tipped minimum wage using w-2

data.” US Census Bureau, May, 26.

Kline, Patrick, Neviana Petkova, Heidi Williams, and Owen Zidar (2019), “Who profts

from patents? rent-sharing at innovative frms.” The quarterly journal of economics, 134,

1343–1404.

Lamadon, Thibaut, Magne Mogstad, and Bradley Setzler (2022), “Imperfect competition,

compensating diferentials, and rent sharing in the us labor market.” American Economic

Review, 112, 169–212.

Neumark, David and Maysen Yen (2023), “The employment and redistributive efects of

reducing or eliminating minimum wage tip credits.” Journal of Policy Analysis and Management, 42, 1092–1116.

Oettinger, Gerald S (1999), “An empirical analysis of the daily labor supply of stadium

vendors.” Journal of political Economy, 107, 360–392.

Schweitzer,

Justin

(2021),

“Ending

the

tipped

minimum

wage

will

reduce

poverty and inequality.” URL https://www.americanprogress.org/article/

ending-tipped-minimum-wage-will-reduce-poverty-inequality/.

Wessels, Walter John (1993), “The minimum wage and tipped employees.” Journal of Labor

Research, 14, 213–226.

Wessels, Walter John (1997), “Minimum wages and tipped servers.” Economic Inquiry, 35,

334–349.

Figures

Figure 1: Prevalence of restaurant employment

(a) Restaurant employment by income decile

(b) Restaurant employment by educational attainment

18.3

Share Employed in Restaurants/Bars (%)

18.1

13.4

9.3

5.8

3.7

2.8

2.1

1.4

0.9

0.8

Income Decile (State-Level, 2023)

7.5

1.7

Less than HS

4 year College

Educational attainment, 2023

Notes: Panel A plots the share of employed workers working in restaurants or bars in each decile of the

income distribution. Deciles are defned at the state-year level. Panel B plots the share of employed

workers working in restaurants or bars by educational attainment. These fgures are made using the

ACS (American Community Survey) 2023 1% sample. In the ACS, the reported industry is generally

“the industry from which the person earned the most money.”

Minimum Wage

Figure 2: 2019 Tipped

Minimum Wage

Notes: This maps the ratio of the tipped minimum wage to the minimum wage in 2019.

Figure 3: Efects of events on log(TMW) and log(MW)

(a) Tipped minimum wage change events

(b) Minimum wage change events

TMW

.15

Log(MW), Log(TMW)

TMW

.15

.05

-.05

-.1

-3

-2

-1

-3

Event Time

-2

-1

Event Time

Note: These plots estimate the efect of tipped minimum wage increase events on the tipped minimum

wage and the minimum wage and the efect of minimum wage increase events on both policies. They

show estimates from equation 2. 95% confdence intervals are shown.

Figure 4: Efect of TMW and MW on composition of earnings

(a) Efect of TMW on types of earnings

.05

log(base wages)

log(tips)

log(earnings)

Coefficient

-.05

-.1

-3

-2

-1

Event Time

(b) Efect of MW on types of earnings

.08

log(base wages)

log(tips)

log(earnings)

Coefficient

.06

.04

.02

-.02

-3

-2

-1

Event Time

.02

TMW

Coefficient

.01

-.01

-.02

-3

-2

-1

Event Time

Note: Panels (a) and (b) estimate the efect of tipped minimum wage increases (TMW) and

minimum wage increases (MW) on reported log(base wages), log(tips), and log(earnings) for

tipped workers. The regressions are run at the frm-state level and are weighted by employment

(see Equation 1). Panel (c) shows the efect of the two policies on the share of earnings that are

tips, and does not limit to full-year workers, unlike panels (a) and (b). Earnings are always the

sum of base wages and tips. Standard errors are clustered at the state level. 95% confdence

intervals are shown.

Figure 5: Efect of TMW and MW on within-frm earnings dispersion

(a) Efect of TMW on base wages

.04

(b) Efect of MW on base wages

.04

Q1 log(base wages)

Q2 log(base wages)

Q3 log(base wages)

Q4 log(base wages)

.02

Coefficient

.02

Q1 log(base wages)

Q2 log(base wages)

Q3 log(base wages)

Q4 log(base wages)

-.02

-3

-2

-1

-3

-2

-1

Event Time

.02

Event Time

(d) Efect of MW on reported earnings

.06

Q1 log(earnings)

Q2 log(earnings)

Q3 log(earnings)

Q4 log(earnings)

Q1 log(earnings)

Q2 log(earnings)

Q3 log(earnings)

Q4 log(earnings)

.04

Coefficient

.02

-.02

-.04

-.02

-3

-2

-1

-3

-2

-1

Event Time

(e) Efect of TMW on reported tips

.05

Event Time

(f) Efect of MW on reported tips

.15

Q1 log(tips)

Q2 log(tips)

Q3 log(tips)

Q4 log(tips)

Q1 log(tips)

Q2 log(tips)

Q3 log(tips)

Q4 log(tips)

Coefficient

.05

-.05

-.1

-.05

-3

-2

-1

-3

Event Time

-2

-1

Event Time

Note: These plots estimate the efect of tipped minimum wage increases (TMW) and minimum wage

increases (MW) on reported log(base wages), log(tips), and log(earnings) for workers in each quartile

of their frm’s earnings distribution. Regressions are run at the frm-state level and are weighted by

employment (see Equation 1). Standard errors are clustered at the state level. 95% confdence intervals

are shown. Quartiles are defned using all full-year workers (included non-tipped workers) for frms with

at least 4 full-year W-2s.

Figure 6: Efect of TMW and MW on employment, log(W-2 count)

Coefficient

.05

TMW

-.05

-3

-2

-1

Event Time

Note: This plot estimates the efect of tipped minimum wage increases (TMW) and minimum wage

increases (MW) on the log(W-2 count). Regressions are run at the frm-state level and are weighted by

employment (see Equation 1). Standard errors are clustered at the state level. 95% confdence intervals

are shown.

Figure 7: Efect of TMW and MW on wage bill, tip bill, and compensation bill

(a) Tipped minimum wage change events

.02

log(wage bill)

log(tip bill)

log(compensation bill)

Coefficient

-.02

-.04

-.06

-.08

-3

-2

-1

Event Time

(b) Minimum wage change events

log(wage bill)

log(tip bill)

log(compensation bill)

Coefficient

.05

-.05

-3

-2

-1

Event Time

Note: These plots estimate the efect of tipped minimum wage increases (TMW) and minimum

wage increases (MW) on log(wage bill), log(tip bill), and log(compensation bill). The wage bill

is the sum of all base wages at the frm, the tip bill is the sum of all tips, and the compensation

bill is the sum of both. There are no restrictions on whether workers earn tips or whether they

are full-year workers. Regressions are run at the frm-state level and are weighted by employment

(see Equation 1). Standard errors are clustered at the state level. 95% confdence intervals are

shown.

Figure 8: Efect of TMW and MW on revenue, tip rate, and share making tips

(a) Efect of TMW and MW on log(revenue)

TMW

.04

Coefficient

.02

-.02

-.04

-.06

-3

-2

-1

Event Time

(b) Efect of TMW and MW on log(tip rate)

TMW

Coefficient

.05

-.05

-3

-2

-1

Event Time

any tips

Coefficient

.005

TMW

-.005

-3

-2

-1

Event Time

Note: This plot estimates the efect of tipped minimum wage increases (TMW) and minimum

wage increases (MW) on the log(revenue), log(Tip Rate)—the ratio of tips to revenue, and the

share of workers making any tips. Regressions are run at the frm-state level and are weighted by

employment (see Equation 1). Standard errors are clustered at the state level. 95% confdence

intervals are shown.

Tables

Table 1: Summary statistics for 2019 restaurants (mean and std. dev.)

(1)

Bars

(2)

Full Service

Restaurants

(3)

Limited Service

Restaurants

Total

26,167

(10.19%)

143,382

(55.85%)

87,159

(33.95%)

256,708

(100.00%)

W-2 Base Wages

8,931.25

(5,572.36)

9,871.83

(5,395.94)

10,257.36

(5,883.82)

9,906.85

(5,596.52)

W-2 Earnings

11,677.42

(6,578.47)

12,611.89

(6,207.89)

10,914.43

(6,077.83)

11,940.31

(6,252.55)

W-2 Tips

5,887.60

(4,685.00)

6,482.54

(4,615.00)

3,473.92

(3,924.60)

5,855.30

(4,648.13)

Share Workers Tipped

0.48

(0.37)

0.45

(0.30)

0.22

(0.34)

0.37

(0.34)

Tips/Earnings (if >0)

0.48

(0.20)

0.50

(0.19)

0.30

(0.21)

0.46

(0.21)

Employment

17.64

(25.41)

39.31

(131.43)

43.90

(231.00)

38.66

(166.99)

1,078,083.82

(1,465,498.78)

1,161,235.39

(1,368,150.31)

Revenue

742,723.34

1,287,558.39

(909,838.08) (1,358,083.51)

(4)

Tips/Revenue

0.06

(0.06)

0.07

(0.06)

0.02

(0.03)

0.05

(0.06)

Labor Share

0.25

(0.14)

0.30

(0.36)

0.26

(0.24)

0.28

(0.31)

Notes: Means are presented, and standard deviations are presented in parentheses. This table provides

summary statistics for all bars, full service restaurants and limited service restaurants in 2019. Column (4)

provides summary statistics for all frms. Observations are frms. Individual level variables are averaged at

the frm-state level, then collapsed to the frm level, weighted by employment.

Table 2: Summary statistics for regression samples (mean and std. dev.)

(1)

Main

(2)

Full Service

(3)

Single State

484,156

366,683

364,312

Base Wages (full year)

13,687.60

(5,004.78)

13,575.23

(4,770.99)

13,713.06

(4,841.54)

Earnings (full year)

19,210.36

(6,399.79)

19,621.14

(6,354.02)

19,384.63

(6,429.25)

Tipped Worker Base Wages (full year)

9,524.06

(5,155.75)

8,967.83

(4,871.93)

9,596.59

(5,047.16)

Tipped Worker Earnings (full year)

18,348.97

(7,051.78)

18,712.28

(7,081.89)

18,590.96

(7,063.60)

Tips (full year)

8,824.91

(5,353.01)

9,744.45

(5,114.81)

8,994.37

(5,356.50)

Tips/Earnings (if >0)

0.48

(0.19)

0.52

(0.16)

0.48

(0.19)

Tips/Earnings

0.28

(0.13)

0.30

(0.11)

0.28

(0.12)

Share Workers Tipped

0.61

(0.19)

0.60

(0.17)

0.62

(0.18)

Employment (W2s)

85.96

(236.89)

83.44

(237.76)

63.53

(81.64)

Revenue

2,503,883.01

2,617,974.59

2,248,359.41

(3,806,238.83) (3,934,781.85) (2,698,197.16)

Tips/Revenue

0.09

(0.04)

0.10

(0.04)

0.09

(0.04)

Tipped Worker Base Wages

6,139.65

(3,506.93)

5,824.43

(3,301.82)

6,191.34

(3,365.08)

Tipped Worker Earnings

11,977.04

(5,150.49)

12,247.12

(5,124.82)

12,143.91

(5,115.81)

Notes: Means are presented, and standard deviations are presented in parentheses. This table provides summary statistics for

our regression samples. Observations span all years used and are at the frm-state level. Column (1) is our main sample, used

for all main results and for the robustness checks using alternative event specifcations. Column (2) is used for the robustness

check on full service restaurants. Column (3) restricts to frms that had at least 80% of their W-2s in a single state in every

year in our sample.

Table 3: Summary statistics for regression samples (mean and std.

dev.)

(1)

Main

(2)

Full Service

(3)

Single State

484,156

366,683

364,312

Wage Bill

747,105.22

670,062.94

528,677.86

(2,343,945.46) (1,853,496.82) (669,994.35)

Tip Bill

244,628.05

(648,552.58)

Compensation Bill

991,733.27

946,496.46

735,805.34

(2,836,484.99) (2,529,486.99) (878,374.82)

276,433.52

(725,942.77)

207,127.48

(252,925.38)

Tips

5,837.39

(3,628.77)

6,422.68

(3,489.28)

5,952.57

(3,636.64)

Hires

37.20

(104.37)

36.35

(106.17)

27.36

(40.54)

Separations

36.63

(101.07)

35.88

(103.29)

26.89

(39.40)

Notes: Means are presented, and standard deviations are presented in parentheses. This table provides summary statistics for

our regression samples. Observations span all years used and are at the frm-state level. Column (1) is our main sample, used

for all main results and for the robustness checks using alternative event specifcations. Column (2) is used for the robustness

check on full service restaurants. Column (3) restricts to frms that had at least 80% of their W-2s in a single state in every

year in our sample.

Table 4: Summary statistics for regression samples (mean and std. dev.)

(1)

Main

(2)

(3)

Full Service Single State

484,156

366,683

364,312

Q1 Tipped Worker Base Wages (full year)

4,720.67

(3,512.28)

4,523.08

(3,371.64)

4,784.72

(3,468.44)

Q2 Tipped Worker Base Wages (full year)

7,729.15

(5,137.92)

7,314.47

(4,927.00)

7,819.30

(5,071.27)

Q3 Tipped Worker Base Wages (full year)

10,359.98

(6,504.95)

9,694.00

(6,211.14)

10,469.76

(6,415.74)

Q4 Tipped Worker Base Wages (full year)

17,044.27

(10,670.22)

15,948.76

(10,298.91)

17,120.39

(10,565.26)

Q1 Tipped Worker Earnings (full year)

8,966.50

(5,114.05)

9,153.97

(5,143.96)

9,095.82

(5,132.96)

Q2 Tipped Worker Earnings (full year)

15,542.84

(7,145.34)

15,910.22

(7,199.94)

15,770.01

(7,176.58)

Q3 Tipped Worker Earnings (full year)

21,401.60

(8,598.44)

21,888.72

(8,643.80)

21,726.08

(8,632.06)

Q4 Tipped Worker Earnings (full year)

30,346.75

(11,015.39)

30,722.61

(10,974.45)

30,725.77

(11,031.07)

Q1 Tips (full year)

4,245.83

(3,404.07)

4,630.89

(3,410.33)

4,311.10

(3,399.01)

Q2 Tips (full year)

7,813.68

(5,545.60)

8,595.75

(5,467.12)

7,950.71

(5,543.85)

Q3 Tips (full year)

11,041.62

(7,183.27)

12,194.72

(6,947.46)

11,256.31

(7,201.83)

Q4 Tips (full year)

13,302.49

(9,133.83)

14,773.85

(8,823.95)

13,605.38

(9,176.41)

Notes: Means are presented, and standard deviations are presented in parentheses. This table provides summary statistics for

our regression samples. Observations span all years used and are at the frm-state level. Column (1) is our main sample, used

for all main results and for the robustness checks using alternative event specifcations. Column (2) is used for the robustness

check on full service restaurants. Column (3) restricts to frms that had at least 80% of their W-2s in a single state in every

year in our sample.

Table 5: Summary statistics for placebo regression samples

(mean and std. dev.)

(1)

5% sample of frms (all industries)

1,176,463

Base Wages (full year)

39,135.89

(26,405.93)

Earnings (full year)

39,528.37

(26,113.28)

Tips/Earnings

0.02

(0.09)

Employment (W2s)

54.43

(332.66)

Wage Bill

1,444,862.18

(18314358.89)

Tip Bill

13,420.61

(138,757.50)

Compensation Bill

1,458,282.79

(18317698.22)

Share Workers Tipped

0.06

(0.18)

Hires

20.97

(160.01)

Separations

19.97

(154.39)

Notes: Means are presented, and standard deviations are presented in parentheses. This table provides

summary statistics for our fnal regression sample. Observations span all years used and are at the frmstate level. Column (1) is our placebo sample, using the 5% sample of frms from all industries rather than

restaurants.

Appendix

Additional Figures

Figure A1: Events

(Tipped) Minimum Wage Changes, 2003–2019

2006

2005

2004

2003

HI HI

MAMA

MNMN

CACA

COCO

DCDC

HI HI

MDMD

MTMT

NDND

NHNH

NVNV

OKOK

OROR

WVWV

WAWA

MI MI

MNMN

MEME

MOMO

NVNV

OROR

RI RI

SDSD

Red= TMW events; Blue= MW events

Note: This fgure shows all events used in our main specifcation.

OROR

WAWA

VT VT

WVWV

2019

MOMO

MTMT

NDND

NHNH

AZ AZ

CACA

COCO

2018

NYNY

OHOH

2017

NYNY

AZ AZ

COCO

CTCT

2009

NYNY

IA IA

IL IL

MDMD

FL FL

2008

MOMO

MTMT

CACA

2016

MDMD

MNMN

HI HI

CACA

2015

IL IL

AKAK

2014

HI HI

IL IL

AKAK

2013

AKAK

2012

FL FL

CACA

ARAR

AZ AZ

2007

2011

AZ AZ

CACA

COCO

2010

Figure A2: Efect of MW and TMW on hires and separations

(a) Efect of TMW on hires and separations

.05

(b) Efect of MW on hires and separations

log(hires)

log(separations)

log(hires)

log(separations)

.08

Coefficient

-.05

-.1

.06

.04

.02

-.15

-3

-2

-1

-3

-2

-1

Event Time

Note: This plot estimates the efect of tipped minimum wage increases (TMW) and minimum wage

increases (MW) on the log(hires) and log(separations). Regressions are run at the frm-state level and

are weighted by employment (see Equation 1). Standard errors are clustered at the state level. 95%

confdence intervals are shown.

Figure A3: Efect of MW and TMW on earnings for all full-year workers (including nontipped workers)

(a) Efect of TMW on earnings

.02

(b) Efect of MW on earnings

log(earnings)

.01

Coefficient

log(earnings)

.015

.005

-.01

-.005

-.02

-3

-2

-1

-3

Event Time

-2

-1

Event Time

Note: This plot estimates the efect of tipped minimum wage increases (TMW) and minimum wage

increases (MW) on the log(earnings), including non tipped workers. Regressions are run at the frmstate level and are weighted by employment (see Equation 1). Standard errors are clustered at the

state level. 95% confdence intervals are shown.

Figure A4: Efect of MW and TMW on earnings for single state frms

(a) Efect of TMW on earnings

.05

(b) Efect of MW on earnings

log(base wages)

log(tips)

log(earnings)

log(base wages)

log(tips)

log(earnings)

.05

Coefficient

-.05

-.1

-.05

-3

-2

-1

-3

-2

-1

Event Time

Note: This plot estimates the efect of tipped minimum wage increases (TMW) and minimum wage

increases (MW) on the log(base wages), log(tips), and log(earnings) of tipped workers. These results

are restricted to frms that had at least 80% of their W-2s in a single state in every year in our sample.

Regressions are run at the frm-state level and are weighted by employment (see Equation 1). Standard

errors are clustered at the state level. 95% confdence intervals are shown.

Figure A5: Efect of MW and TMW on earnings for all tipped workers (including non-fullyear)

(a) Efect of TMW on earnings

(b) Efect of MW on earnings

log(base wages)

log(tips)

log(earnings)

.06

.04

Coefficient

.04

Coefficient

log(base wages)

log(tips)

log(earnings)

.06

.02

-.02

.02

-.02

-.04

-3

-2

-1

-3

Event Time

-2

-1

Event Time

Note: This plot estimates the efect of tipped minimum wage increases (TMW) and minimum wage

increases (MW) on the log(base wages), log(tips), and log(earnings) of tipped workers. These results

include full-year and non-full-year workers. Regressions are run at the frm-state level and are weighted

by employment (see Equation 1). Standard errors are clustered at the state level. 95% confdence

intervals are shown.

Figure A6: Efect of MW and TMW on earnings for full-service restaurants

(a) Efect of TMW on earnings

(b) Efect of MW on earnings

.06

log(base wages)

log(tips)

log(earnings)

.05

log(base wages)

log(tips)

log(earnings)

Coefficient

.04

.02

-.02

-.05

-.04

-3

-2

-1

-3

-2

-1

Event Time

Note: This plot estimates the efect of tipped minimum wage increases (TMW) and minimum wage

increases (MW) on the log(base wages), log(tips), and log(earnings) of tipped workers. These results

are restricted to frms that are ever classifed as full-service restaurants. Regressions are run at the

frm-state level and are weighted by employment (see Equation 1). Standard errors are clustered at

the state level. 95% confdence intervals are shown.

Figure A7: Efect of MW and TMW on earnings excluding federal events

(a) Efect of TMW on earnings

.05

(b) Efect of MW on earnings

log(base wages)

log(tips)

log(earnings)

Coefficient

.05

Coefficient

log(base wages)

log(tips)

log(earnings)

-.05

-.1

-.05

-3

-2

-1

-3

Event Time

-2

-1

Event Time

Note: This plot estimates the efect of tipped minimum wage increases (TMW) and minimum wage

increases (MW) on the log(base wages), log(tips), and log(earnings) of tipped workers. These regressions

exclude federal changes as events. Regressions are run at the frm-state level and are weighted by

employment (see Equation 1). Standard errors are clustered at the state level. 95% confdence intervals

are shown.

Figure A8: Efect of MW and TMW on earnings collapsing consecutive events

(a) Efect of TMW on earnings

(b) Efect of MW on earnings

.15

log(base wages)

log(tips)

log(earnings)

Coefficient

.05

log(base wages)

log(tips)

log(earnings)

.05

-.05

-.1

-.05

-3

-2

-1

-3

-2

-1

Event Time

Note: This plot estimates the efect of tipped minimum wage increases (TMW) and minimum wage

increases (MW) on the log(base wages), log(tips), and log(earnings) of tipped workers. These results

use a defnition of event where multi-year increases are collapsed to the frst year. Regressions are run

at the frm-state level and are weighted by employment (see Equation 1). Standard errors are clustered

at the state level. 95% confdence intervals are shown.

Figure A9: Efect of MW and TMW on earnings for 5% sample of all frms (all industries)

(a) Efect of TMW on earnings

.02

(b) Efect of MW on earnings

.015

log(base wages)

log(earnings)

log(base wages)

log(earnings)

.01

Coefficient

.01

.005

-.01

-.005

-.02

-.01

-3

-2

-1

-3

Event Time

-2

-1

Event Time

Note: This plot estimates the efect of tipped minimum wage increases (TMW) and minimum wage

increases (MW) on the log(base wages), log(tips), and log(earnings) of tipped workers. These results

use the 5% sample of all frms rather than the restaurant sample. Regressions are run at the frm-state

level and are weighted by employment (see Equation 1). Standard errors are clustered at the state

level. 95% confdence intervals are shown.

Proofs

Proof of Proposition 1:

The frm’s problem when subject to the tipped minimum wage w̄ is:

max P (T )L(w

¯ + T ) − wL(

¯ w¯ + T )

First order condition:

−P ′ (T )L(T + w̄) = (P (T ) − w̄)L′ (T + w̄)

Implicitly diferentiating this expression 17 :

− (P ′′ (T )L + P ′ (T )L′ ) dT − P ′ (T )L′ dw̄

= (P ′ (T )L′ + (P (T ) − w)L

¯ ′′ ) dT + (−L′ + (P (T ) − w)L

¯ ′′ ) dw

Rearranging:

−P ′ (T )L′ + L′ + (P (T ) − w)L

¯ ′′

dw̄

2P ′ (T )L′ + (P (T ) − w)L

¯ ′′ + P ′′ (T )L

Firms weakly proft on their workers so P (T ) − w̄ ≥ 0. This allows us to sign each term

on the numerator as positive and each term on the denominator as negative.

Of course, tips cannot fall negative—so if tips are already zero then they cannot fall

further. So tips weakly fall in response to increases in the binding tipped minimum wage.

Proof of Proposition 2:

was above

To know whether total earnings increased or decreased, we want to know if dw̄

or below −1 (the point at which the positive base wage efect is fully cancelled out by a

negative tip efect). For this proposition, we want to know the efect of a small increase to

the tipped minimum wage above the unregulated wage. To answer this, we will now solve

the unregulated problem.

max P (T )L(w + T ) − wL(w + T )

T,w

The wage FOC is P (T )L′ = L+wL′ . The tip FOC is P ′ (T )L+P (T )L′ = wL′ . Combining

these two expressions, we have −P ′ (T ) = 1. In the unregulated equilibrium, the frm trades

of tip and base wage pay. Holding constant the frm’s employment, the cost of paying a

dollar of base wage pay (per worker) is 1 while the cost of a dollar of tip pay (per worker) is

−P ′ (T )—the extent to which prices fall.

Let L(T + w̄), L′ (T + w̄), L′′ (T + w̄) be denoted by L, L′ , L′′ for simplicity

Plugging in the expression, −P ′ (T ) = 1, which holds in the unregulated equilibrium to

from the previous proof, we have:

the expression for dw̄

(P (T ) − w)L

¯ ′′

(w̄ = wunreg ) =

dw̄

−2L′ + (P (T ) − w)L

¯ ′′ + P ′′ (T )L

Since the denominator is larger in magnitude than the numerator, dw̄

(w̄ = wunreg ) > −1.

Proof of Proposition 3:

Now, we ask whether it is possible for an increase to the tipped minimum wage to cause

(w̄) < −1 when P (T )

earnings to fall. We will show that around the value of T = 0, dw̄

has low curvature for small tip values. This which suggests that there exists some threshold

(w̄ = wcutof f ) = −1.

value of tipped minimum wage which causes a level of tips where dw̄

First, note that there will be some value of the tipped minimum wage at which T = 0.

If w̄ = P (0), for example, the frm would lose money on each worker if they paid any tips.

′ (T )L′ +L′ +(P (T )−w)L

¯ ′′

From the proof of proposition X, we have dw̄

= 2P−P

. This expression

′ (T )L′ +(P (T )−w)L

¯ ′′ +P ′′ (T )L

is less than −1 if:

(1 + P ′ (T ))L′ > −P ′′ (T )L

Around T = 0, we have:

L′ > −P ′′ (0)L

< −1. In particular, many

If P ′′ (0) is sufciently small, then this inequality holds and dw̄

′′

reasonable function forms will have P (0) = 0 which unambiguously satisfes the inequality.

This is a copy of a public record, reproduced as it was published. It is not legal advice, and it may not be the version a court would rely on. Check the official source before you cite it.