Fast-food jobs after New Jersey's 1992 minimum-wage rise

1 The gist · for everyone

When New Jersey raised its minimum wage in April 1992, did fast-food restaurants there cut jobs compared with similar restaurants just across the border in Pennsylvania, where the wage did not change?

+2.75

more workers per New Jersey restaurant than the Pennsylvania trend predicts

Plausible range +0.09 to +5.42

We counted staff at fast-food restaurants in early 1992 and again late in the year. New Jersey raised its on 1 April; nearby Pennsylvania did not. We compare how the average staff per restaurant changed in each state.

New Jersey restaurants averaged 20.4 full-time-equivalent workers before the rise, so this is about 13% of a typical store.

Picture an average New Jersey restaurant with about 20 full-time-equivalent workers. If it had followed Pennsylvania's path it would have ended the year with about 18; it ended with about 21.

What kind of claim: Points toward a cause, with assumptions. The design removes some alternative explanations, not all. Read it as a cause only if its assumptions hold.

Take New Jersey's change in average staff between the two rounds. Subtract Pennsylvania's change over the same months. What is left is the change Pennsylvania's path does not explain. Dashed line: where New Jersey would have ended if it had changed as much as Pennsylvania. The axis does not start at zero.

Safe to say

"Employment in New Jersey fast-food restaurants did not fall relative to Pennsylvania after the 1992 rise; it grew by about 2.75 full-time-equivalent workers per store more."

Not this

"Raising the minimum wage creates jobs," or "New Jersey restaurants hired 2.75 more people."

Before you repeat it

Pennsylvania staffing fell over the year, so most of the gap comes from Pennsylvania shrinking, not New Jersey growing. The design assumes both states would otherwise have moved alike, and the interval's width depends on a calculation the paper does not describe.

2Who exactly is counted, and when?Follow one restaurant through the rules 3What could make this misleading?4 caveats · 3 common misreadings 4How would I rebuild it?Exact rules, data, code · 5 doc-vs-code gaps

2 How it's built · for the curious

How the number is made

Follow one restaurant: A Wendy's in Camden County

A hypothetical New Jersey Wendy's paying the old $4.25 starting wage has 8 full-time staff, 2 managers and 20 part-timers in February 1992.

1day -40 First survey call: 8 + 2 + 20 × 0.5 = 20 workers
2day 0 rises to $5.05
3day 225 Second survey: 9 full-time, 2 managers, 22 part-time = 22 counts

Result: The restaurant adds +2 to New Jersey's change. If it had shut for renovation in November, it would drop out of the second-round average entirely. If it had closed for good, it would count as 0 workers.

Now the general rule, one step at a time. Each step has the plain version, with the exact version one click away.

Step 1 · Who is counted

Start with restaurants, then apply each rule

One fast-food restaurant from one of four large burger or chicken chains.

Exactly

One row per store in public.dat (410 rows); CHAIN 1 = Burger King, 2 = KFC, 3 = Roy Rogers, 4 = Wendy's. SHEET is documented as the unique store id, but sheet 407 is used by two different stores; rows, not SHEET values, are the unit.
S3S6

All restaurants

Keep if Fast-food restaurants in New Jersey and eastern Pennsylvania from four chains that took part in the first telephone survey.410 remain

Keep if Reported staff numbers in that survey round. Each round's average uses every restaurant with a count for that round, so the two rounds are not exactly the same restaurants.

Keep if Restaurants that closed for good by the second round count as having zero staff. The four that were closed only temporarily are left out of the second round.

Counted in the denominator
Step 2 · When the clock starts

Time zero: wage rise

New Jersey's minimum wage went from $4.25 to $5.05 an hour on 1 April 1992.
- First survey round: The telephone survey from 15 February to 4 March 1992, before the rise. (days −46 to −28)
- Second survey round: The follow-up survey from 5 November to 31 December 1992, about 7 to 9 months after the rise. (days 218–274)
Exactly

1992-04-01 (paper, introduction and section I). Pennsylvania stayed at the federal $4.25.
S1
Step 3 · What counts

Staff after

The same count, taken in the second survey round.

Also measured the same way:
- Staff before: Full-time workers, including managers, plus half the number of part-time workers.
Exactly

EMPTOT2 = EMPPT2*.5 + EMPFT2 + NMGRS2 (check.sas line 55).
S2
Step 4 · Combine

Average across people

Average the staff count over restaurants in each state and survey round.

Exactly

PROC MEANS of EMPTOT and EMPTOT2 BY NJ (check.sas lines 128–131, 'PART OF TABLE 3').
S2S5

total full-time-equivalent workers across restaurants with a count

restaurants with a count in that round319

= 21.03New Jersey, Nov–Dec
Step 5 · Compare

Subtract the usual change

Take New Jersey's change in average staff between the two rounds. Subtract Pennsylvania's change over the same months. What is left is the change Pennsylvania's path does not explain.
(NJ₂ − NJ₁) − (PA₂ − PA₁)
With the real numbers: (21.0274 − 20.4394) − (21.1656 − 23.3312) = 2.754
Whose effect? Effect among those who actually did it (not everyone).

Exactly

(NJ wave 2 − NJ wave 1) − (PA wave 2 − PA wave 1), FTE per store (Table 3, row 3, column iii).
S1S5
Step 6 · How sure

Show the range of plausible values

The paper reports a of 2.03 for this number. We turned that into a of 0.09 to 5.42 workers. The paper does not say how it computed the spread.
Not covered by this range: The calculation behind the published spread is not stated, and check.sas does not compute it. Restaurants appear in both rounds, so the four averages are not independent; the simplest formula that ignores this gives a range that includes zero. Nothing here covers the assumption that both states would have moved alike.

Exactly

SE = 2.76 / 2.03 ≈ 1.36 (paper, section III.A); interval = 2.754 ± 1.96 × 1.36. Treating the four state-by-round means as independent gives SE 1.80 (interval −0.78 to 6.28). The balanced-sample change (row 4) gives SE 1.34.
S1S5

Each dot is one of 20 equally likely values, given sampling noise alone (normal approximation to the interval).

The same definition, assembled mechanically from the spec

Generated from the structured definition, not written by hand. If it disagrees with the summary above, the summary is wrong.

The unit is one restaurant. One fast-food restaurant from one of four large burger or chicken chains.
Keep restaurants that meet every rule, in order: fast-food restaurants in New Jersey and eastern Pennsylvania from four chains that took part in the first telephone survey; reported staff numbers in that survey round. Each round's average uses every restaurant with a count for that round, so the two rounds are not exactly the same restaurants; restaurants that closed for good by the second round count as having zero staff. The four that were closed only temporarily are left out of the second round.
Time zero is the wage rise: New Jersey's minimum wage went from $4.25 to $5.05 an hour on 1 April 1992.
First survey round (observation): the 19 days up to before the wage rise (days -46 to -28).
Second survey round (outcome): days 218–274 after the wage rise.
For each restaurant, record staff before as a value within the first survey round: Full-time workers, including managers, plus half the number of part-time workers.
For each restaurant, record staff after (headline) as a value within the second survey round: The same count, taken in the second survey round.
Average the staff count over restaurants in each state and survey round.
Do this separately for each group: Pennsylvania, Feb–Mar, Pennsylvania, Nov–Dec, New Jersey, Feb–Mar, New Jersey, Nov–Dec.
Take New Jersey's change in average staff between the two rounds. Subtract Pennsylvania's change over the same months. What is left is the change Pennsylvania's path does not explain.
The paper reports a t statistic of 2.03 for this number. We turned that into a 95% range of 0.09 to 5.42 workers. The paper does not say how it computed the spread.

3 Keep in mind · before you act on it

What could make this misleading

Complete. The latest wage rise is 01 Apr 1992; its second survey round ends 31 Dec 1992. Data are complete through 31 Dec 1992, so every window is fully observed. This number should not move as more data arrive.

Caveats

↑ May make the number look bigger than it really is↓ May make the number look smaller than it really is± The true uncertainty is larger than shown↕ Could push the number either way◎ Limits what the number can be used to conclude

◎

Pennsylvania fell more than New Jersey rose

New Jersey's average went up by 0.59 workers while Pennsylvania's fell by 2.17. Most of the gap is Pennsylvania shrinking. The paper attributes this to a worsening regional economy, which the design assumes hit both states equally.

Limits what the number can be used to conclude · high importance S1 S5

±

The spread depends on an unstated calculation

The published implies a of about 0.1 to 5.4 workers. A simple calculation that treats the two rounds as separate samples gives −0.8 to 6.3, which includes no effect. Counting only restaurants seen in both rounds gives 0.1 to 5.4 again.

The true uncertainty is larger than shown · medium importance S1 S5

±

Self-reported counts from a phone survey

Managers reported staff numbers by phone. The paper estimates the reliability of this measure at 0.70 by reinterviewing a subset of stores, so some changes are noise.

The true uncertainty is larger than shown · medium importance S1

◎

One region, one industry, one rise

This is four fast-food chains in New Jersey and eastern Pennsylvania in 1992, after an 80-cent rise. It says nothing directly about other industries, larger rises or other places.

Limits what the number can be used to conclude · medium importance

Common misreadings

Tempting, but wrong	Accurate
Each New Jersey restaurant hired 2.75 more people.	It is 2.75 more workers on average, relative to Pennsylvania's trend. Part-timers count as half, and New Jersey's own average rose by only 0.59.
This proves minimum-wage rises create jobs.	It shows employment did not fall relative to the comparison state in this setting. The design rests on both states sharing a trend, and the result is specific to this case.
Employment rose 13%.	The gap is 13% of a typical New Jersey store's starting staff. New Jersey's own average rose about 3%.

Same question, other reasonable choices

Each row reruns the analysis with one choice changed. If the rows disagree a lot, the headline depends on that choice.

Same restaurants in both rounds (row 4): 384 restaurants with a count in both rounds; SE 1.34.
Temporarily closed counted as zero (row 5): Paper: 'only slightly smaller'.
Regression, Table 4 sample: 357 restaurants with wage data in both rounds; coefficient 2.33 (SE 1.19).

Judgment calls

Comparison group

Chose: Restaurants in eastern Pennsylvania, near the border.

Could have chosen: New Jersey restaurants already paying $5.00 or more, who were barely affected.

Why: Similar markets and seasons, with no wage change.

What it changes: The paper reports both. High-wage New Jersey stores changed by −2.16, almost the same as Pennsylvania's −2.28 (balanced sample), which supports the Pennsylvania comparison.

Part-time workers count as half

Chose: Full-time equivalent = full-time + managers + 0.5 × part-time.

Could have chosen: Count heads; weight part-timers 0.4 or 0.6; exclude managers.

Why: A simple, stated convention.

What it changes: Table 5 rows 3–5 report little effect on the regression for the change in employment when managers are dropped or part-timers weighted 0.4 or 0.6; not re-run here.

Which restaurants enter each average

Chose: Every restaurant with a count in that round (rows 1–3).

Could have chosen: Only restaurants counted in both rounds (row 4).

Why: Uses all the data.

What it changes: The headline barely moves (2.754 vs 2.750).

4 Reproduce it · for specialists

The exact specification

Download the spec (YAML) Methods section

schema: metric-card/1
id: minimum-wage-did
title: Fast-food jobs after New Jersey's 1992 minimum-wage rise
subtitle: Difference-in-differences in average full-time-equivalent employment per store, New Jersey vs eastern Pennsylvania
status: provisional
owner: data-visual-explainer worked example (public data)
updated: "2026-10-07"
question: >
  When New Jersey raised its minimum wage in April 1992, did fast-food restaurants there cut
  jobs compared with similar restaurants just across the border in Pennsylvania, where the wage
  did not change?

gist:
  sentence: >
    We counted staff at fast-food restaurants in early 1992 and again late in the year. New
    Jersey raised its minimum wage on 1 April; nearby Pennsylvania did not. We compare how the
    average staff per restaurant changed in each state.
  headline:
    value: 2.75
    low: 0.09
    high: 5.42
    unit: other
    decimals: 2
    label: more full-time-equivalent workers per New Jersey restaurant than the Pennsylvania trend predicts
    context: "New Jersey restaurants averaged 20.4 full-time-equivalent workers before the rise, so this is about 13% of a typical store."
  frequency: >
    Picture an average New Jersey restaurant with about 20 full-time-equivalent workers. If it
    had followed Pennsylvania's path it would have ended the year with about 18; it ended with
    about 21.
  reading:
    direction: neutral
    say: >
      "Employment in New Jersey fast-food restaurants did not fall relative to Pennsylvania after
      the 1992 rise; it grew by about 2.75 full-time-equivalent workers per store more."
    dont_say: >
      "Raising the minimum wage creates jobs," or "New Jersey restaurants hired 2.75 more people."
  key_caveat: >
    Pennsylvania staffing fell over the year, so most of the gap comes from Pennsylvania
    shrinking, not New Jersey growing. The design assumes both states would otherwise have moved
    alike, and the interval's width depends on a calculation the paper does not describe.
  claim_strength: quasi-causal

grammar:
  unit:
    singular: restaurant
    plural: restaurants
    plain: One fast-food restaurant from one of four large burger or chicken chains.
    exact: >
      One row per store in public.dat (410 rows); CHAIN 1 = Burger King, 2 = KFC, 3 = Roy Rogers,
      4 = Wendy's. SHEET is documented as the unique store id, but
      sheet 407 is used by two different stores; rows, not SHEET values, are the unit.
    src: [S3, S6]
  population:
    - id: frame
      plain: >
        Fast-food restaurants in New Jersey and eastern Pennsylvania from four chains that took
        part in the first telephone survey.
      exact: "public.dat: 410 stores, STATE = 1 (NJ, 331) or 0 (PA, 79)."
      n_after: 410
      src: [S1, S3]
    - id: has_staff_count
      plain: >
        Reported staff numbers in that survey round. Each round's average uses every restaurant
        with a count for that round, so the two rounds are not exactly the same restaurants.
      exact: >
        Wave-1 mean over stores with non-missing fte1 (PA 77, NJ 321); wave-2 mean over stores
        with non-missing fte2 (PA 77, NJ 319). "All available observations" (Table 3 rows 1–2).
      src: [S1, S5]
    - id: closures
      plain: >
        Restaurants that closed for good by the second round count as having zero staff. The four
        that were closed only temporarily are left out of the second round.
      exact: >
        STATUS2 = 3 (6 stores): wave-2 employment fields are already 0 in public.dat. STATUS2 in
        (2, 4, 5) (4 stores, all NJ): wave-2 fields missing, so excluded from the wave-2 mean.
      src: [S1, S3, S6]
  anchor:
    kind: calendar
    label: Wage rise
    plain: New Jersey's minimum wage went from $4.25 to $5.05 an hour on 1 April 1992.
    exact: "1992-04-01 (paper, introduction and section I). Pennsylvania stayed at the federal $4.25."
    src: S1
  windows:
    - id: wave1
      label: First survey round
      start: -46
      end: -27
      unit: day
      role: observation
      plain: The telephone survey from 15 February to 4 March 1992, before the rise.
      exact: "Wave 1: February 15 – March 4, 1992 (Table 1)."
      src: S1
    - id: wave2
      label: Second survey round
      start: 218
      end: 275
      unit: day
      role: outcome
      plain: The follow-up survey from 5 November to 31 December 1992, about 7 to 9 months after the rise.
      exact: "Wave 2: November 5 – December 31, 1992 (Table 1); DATE2 in the file spans 110592–123192."
      src: [S1, S6]
  measures:
    - id: fte_before
      label: Staff before
      kind: value
      window: wave1
      plain: Full-time workers, including managers, plus half the number of part-time workers.
      exact: "EMPTOT = EMPPT*.5 + EMPFT + NMGRS (check.sas line 54)."
      src: [S2, S1]
    - id: fte_after
      label: Staff after
      kind: value
      window: wave2
      primary: true
      plain: The same count, taken in the second survey round.
      exact: "EMPTOT2 = EMPPT2*.5 + EMPFT2 + NMGRS2 (check.sas line 55)."
      src: S2
  aggregate:
    kind: mean
    measure: fte_after
    numerator: total full-time-equivalent workers across restaurants with a count
    denominator: restaurants with a count in that round
    plain: Average the staff count over restaurants in each state and survey round.
    exact: "PROC MEANS of EMPTOT and EMPTOT2 BY NJ (check.sas lines 128–131, 'PART OF TABLE 3')."
    src: [S2, S5]
  groups:
    - id: pa_pre
      label: Pennsylvania, Feb–Mar
      role: baseline
      period: pre
      year: "1992"
      series: Pennsylvania
      plain: Pennsylvania restaurants in the first round.
      exact: "STATE = 0, non-missing EMPTOT (n = 77)."
    - id: pa_post
      label: Pennsylvania, Nov–Dec
      role: baseline
      period: post
      year: "1992"
      series: Pennsylvania
      plain: Pennsylvania restaurants in the second round.
      exact: "STATE = 0, non-missing EMPTOT2 (n = 77)."
    - id: nj_pre
      label: New Jersey, Feb–Mar
      role: focal
      period: pre
      year: "1992"
      series: New Jersey
      plain: New Jersey restaurants in the first round, before the rise.
      exact: "STATE = 1, non-missing EMPTOT (n = 321)."
    - id: nj_post
      label: New Jersey, Nov–Dec
      role: focal
      period: post
      year: "1992"
      series: New Jersey
      plain: New Jersey restaurants in the second round, after the rise.
      exact: "STATE = 1, non-missing EMPTOT2 (n = 319)."
  comparison:
    kind: difference_in_differences
    target: ATT
    plain: >
      Take New Jersey's change in average staff between the two rounds. Subtract Pennsylvania's
      change over the same months. What is left is the change Pennsylvania's path does not
      explain.
    exact: "(NJ wave 2 − NJ wave 1) − (PA wave 2 − PA wave 1), FTE per store (Table 3, row 3, column iii)."
    formula: "(NJ₂ − NJ₁) − (PA₂ − PA₁)"
    src: [S1, S5]
  uncertainty:
    method: As published (t statistic), method not stated
    level: 0.95
    plain: >
      The paper reports a t statistic of 2.03 for this number. We turned that into a 95% range
      of 0.09 to 5.42 workers. The paper does not say how it computed the spread.
    exact: >
      SE = 2.76 / 2.03 ≈ 1.36 (paper, section III.A); interval = 2.754 ± 1.96 × 1.36. Treating
      the four state-by-round means as independent gives SE 1.80 (interval −0.78 to 6.28). The
      balanced-sample change (row 4) gives SE 1.34.
    known_gaps: >
      The calculation behind the published spread is not stated, and check.sas does not compute
      it. Restaurants appear in both rounds, so the four averages are not independent; the
      simplest formula that ignores this gives a range that includes zero. Nothing here covers
      the assumption that both states would have moved alike.
    src: [S1, S5]

results:
  as_of: "1994-09 (published); recomputed 2026-10-07"
  unit: other
  latest_anchor: "1992-04-01"
  data_complete_through: "1992-12-31"
  cells:
    - {group: pa_pre, value: 23.3312, n: 77}
    - {group: pa_post, value: 21.1656, n: 77}
    - {group: nj_pre, value: 20.4394, n: 321}
    - {group: nj_post, value: 21.0274, n: 319}
  estimate: {value: 2.75, low: 0.09, high: 5.42}
  sensitivity:
    - {label: "Same restaurants in both rounds (row 4)", value: 2.75, low: 0.12, high: 5.38, note: "384 restaurants with a count in both rounds; SE 1.34.", src: S5}
    - {label: "All restaurants with a count (row 3, headline)", value: 2.75, low: 0.09, high: 5.42, primary: true, src: [S1, S5]}
    - {label: "Temporarily closed counted as zero (row 5)", value: 2.51, low: -0.13, high: 5.15, note: "Paper: 'only slightly smaller'.", src: S5}
    - {label: "Regression, Table 4 sample", value: 2.33, low: 0.0, high: 4.66, note: "357 restaurants with wage data in both rounds; coefficient 2.33 (SE 1.19).", src: [S1, S5]}

example:
  name: A Wendy's in Camden County
  intro: >
    A hypothetical New Jersey Wendy's paying the old $4.25 starting wage has 8 full-time staff,
    2 managers and 20 part-timers in February 1992.
  anchor_date: "1992-04-01"
  events:
    - {at: -40, label: "First survey call: 8 + 2 + 20 × 0.5 = 20 full-time-equivalent workers"}
    - {at: 0, label: "Minimum wage rises to $5.05"}
    - {at: 225, label: "Second survey: 9 full-time, 2 managers, 22 part-time = 22", counts: true}
  outcome: >
    The restaurant adds +2 to New Jersey's change. If it had shut for renovation in November, it
    would drop out of the second-round average entirely. If it had closed for good, it would
    count as 0 workers.

fine_print:
  judgment_calls:
    - decision: Comparison group
      choice: Restaurants in eastern Pennsylvania, near the border.
      alternatives: New Jersey restaurants already paying $5.00 or more, who were barely affected.
      why: Similar markets and seasons, with no wage change.
      impact: >
        The paper reports both. High-wage New Jersey stores changed by −2.16, almost the same as
        Pennsylvania's −2.28 (balanced sample), which supports the Pennsylvania comparison.
      src: [S1, S5]
    - decision: Part-time workers count as half
      choice: Full-time equivalent = full-time + managers + 0.5 × part-time.
      alternatives: Count heads; weight part-timers 0.4 or 0.6; exclude managers.
      why: A simple, stated convention.
      impact: >
        Table 5 rows 3–5 report little effect on the regression for the change in employment
        when managers are dropped or part-timers weighted 0.4 or 0.6; not re-run here.
      src: [S1, S2]
    - decision: Which restaurants enter each average
      choice: Every restaurant with a count in that round (rows 1–3).
      alternatives: Only restaurants counted in both rounds (row 4).
      why: Uses all the data.
      impact: The headline barely moves (2.754 vs 2.750).
      src: S5
  caveats:
    - title: Pennsylvania fell more than New Jersey rose
      text: >
        New Jersey's average went up by 0.59 workers while Pennsylvania's fell by 2.17. Most of
        the gap is Pennsylvania shrinking. The paper attributes this to a worsening regional
        economy, which the design assumes hit both states equally.
      effect: limits_scope
      severity: high
      src: [S1, S5]
    - title: The spread depends on an unstated calculation
      text: >
        The published t statistic implies a 95% range of about 0.1 to 5.4 workers. A simple
        calculation that treats the two rounds as separate samples gives −0.8 to 6.3, which
        includes no effect. Counting only restaurants seen in both rounds gives 0.1 to 5.4 again.
      effect: wider_uncertainty
      severity: medium
      src: [S1, S5]
    - title: Self-reported counts from a phone survey
      text: >
        Managers reported staff numbers by phone. The paper estimates the reliability of this
        measure at 0.70 by reinterviewing a subset of stores, so some changes are noise.
      effect: wider_uncertainty
      severity: medium
      src: S1
    - title: One region, one industry, one rise
      text: >
        This is four fast-food chains in New Jersey and eastern Pennsylvania in 1992, after an
        80-cent rise. It says nothing directly about other industries, larger rises or other places.
      effect: limits_scope
      severity: medium
  misreadings:
    - wrong: "Each New Jersey restaurant hired 2.75 more people."
      right: >
        It is 2.75 more full-time-equivalent workers on average, relative to Pennsylvania's trend.
        Part-timers count as half, and New Jersey's own average rose by only 0.59.
    - wrong: "This proves minimum-wage rises create jobs."
      right: >
        It shows employment did not fall relative to the comparison state in this setting. The
        design rests on both states sharing a trend, and the result is specific to this case.
    - wrong: "Employment rose 13%."
      right: >
        The gap is 13% of a typical New Jersey store's starting staff. New Jersey's own average
        rose about 3%.
  discrepancies:
    - topic: Store identifier
      described: "Codebook: SHEET is the 'sheet number (unique store id)'."
      implemented: "Sheet 407 appears twice in public.dat: a Pennsylvania KFC and a New Jersey Wendy's."
      impact: >
        Anyone deduplicating or joining on SHEET loses or mixes one store. The paper's
        averages, which treat rows as stores, are unaffected.
      src: [S3, S5]
    - topic: Where closed stores become zero
      described: "Table 3 note: 'Employment at six closed stores is set to zero.'"
      implemented: >
        check.sas never sets anything to zero; the six STATUS2 = 3 rows already carry 0 in the
        wave-2 employment fields of public.dat.
      impact: >
        None for the published numbers. A re-analysis that recodes STATUS2 itself, or treats 0
        as missing, will silently change the headline.
      src: [S1, S2, S5]
    - topic: Headline value
      described: "Text: 'The relative gain … is 2.76 FTE employees (or 13 percent), with a t statistic of 2.03.'"
      implemented: >
        The public data give 2.754, which rounds to 2.75. Rows 4 and 5 and Table 4 reproduce to
        the published decimals.
      impact: Rounding only. It is likely the text was computed from rounded table entries.
      src: [S1, S5]
    - topic: How the headline's uncertainty was computed
      described: "A t statistic of 2.03; no method given."
      implemented: >
        check.sas computes only per-state means (PROC MEANS), not the standard error of the
        difference. The independent-samples formula gives t = 1.53; the balanced-sample formula
        gives t = 2.05.
      impact: Unknown. It decides whether the headline is distinguishable from zero.
      src: [S1, S2, S5]
    - topic: Restaurants without a starting wage
      described: "Table 3 notes give New Jersey wage-group counts of 101, 140 and 73."
      implemented: >
        Those sum to 314 of 331 New Jersey stores. The 17 with no wave-1 starting wage go to
        check.sas's 'NJ STORE, BAD WAGE' group and are not mentioned in the text.
      impact: Affects only the within-New Jersey columns (iv)–(viii), not the headline.
      src: [S1, S2, S5]

spec:
  estimand:
    population: >
      Burger King, KFC, Wendy's and Roy Rogers restaurants in New Jersey and eastern Pennsylvania
      in the 1992 sample frame that responded in wave 1 and reported employment in the given wave.
    exposure: Being in New Jersey when its minimum wage rose from $4.25 to $5.05 on 1992-04-01.
    time_zero: 1992-04-01.
    variable: >
      FTE employment per store = full-time + managers + 0.5 × part-time, in wave 2 (Nov 5 – Dec 31,
      1992) minus wave 1 (Feb 15 – Mar 4, 1992).
    intercurrent_events: >
      Permanent closures count as zero employment; temporary closures (4) are missing in wave 2.
    summary: >
      Difference in state mean changes (unbalanced means), FTE workers per store; interval from
      the published t statistic.
  parameters:
    - {name: old_minimum, value: "$4.25", meaning: "Federal and NJ rate before 1992-04-01"}
    - {name: new_minimum, value: "$5.05", meaning: "NJ rate from 1992-04-01"}
    - {name: part_time_weight, value: 0.5, meaning: Each part-time worker counts as half}
    - {name: temporarily_closed_status, value: "2, 4, 5", meaning: "STATUS2 codes treated as missing in wave 2"}
  lineage:
    - name: public.dat
      role: One row per store, both survey waves
      grain: store
      src: [S3, S6]
    - name: check.sas
      role: Authors' program; computes FTE and the per-state means of Table 3
      grain: store → state × wave
      src: S2
    - name: examples/minimum-wage/analysis.py
      role: Python reproduction of Table 3 rows 1–5 and Table 4 column (i)
      grain: state × wave
      src: S5
  code:
    - purpose: FTE employment in each wave (authors' SAS)
      lang: sas
      src: S2
      excerpt: |
        EMPTOT=EMPPT*.5 + EMPFT + NMGRS;
        EMPTOT2=EMPPT2*.5 + EMPFT2 + NMGRS2;
        DEMP=EMPTOT2-EMPTOT;;
    - purpose: Difference-in-differences and the independent-samples standard error (reproduction)
      lang: python
      src: S5
      excerpt: |
        did = (lv["NJ_post"]["value"] - lv["NJ_pre"]["value"]) - (lv["PA_post"]["value"] - lv["PA_pre"]["value"])
        se_indep = np.sqrt(sum(r["se"] ** 2 for r in rows))
  reproduce:
    steps:
      - "python examples/fetch_data.py minimum-wage (downloads and checks the data and paper)."
      - "python examples/minimum-wage/analysis.py (writes results/table3.csv and results/checks.csv)."
    receipts:
      - "2026-10-07: PA 23.33 → 21.17, NJ 20.44 → 21.03; DiD 2.754. Row 4 2.750 (SE 1.342), row 5 2.509."
      - "2026-10-07: Table 4 column (i) 2.326 (SE 1.192), n = 357, mean change −0.238, SD 8.825; the paper prints 2.33 (1.19), −0.237, 8.825."
    tests: "examples/minimum-wage/results/checks.csv compares each reproduced figure with the paper."

glossary:
  - term: full-time equivalent
    aliases: [FTE, full-time-equivalent]
    plain: >
      A staff count where each part-time worker counts as half a full-time worker. Managers count
      as full-time.
  - term: difference-in-differences
    aliases: [DiD]
    plain: >
      Compare one group's change with another group's change over the same time. The difference
      between the two changes is the part a shared trend does not explain.
  - term: minimum wage
    plain: The lowest hourly pay an employer may legally offer.
  - term: 95% range
    aliases: [95% interval, confidence interval]
    plain: >
      The values that fit the data well, given random noise. It does not cover the assumption
      that both states would have moved alike.
  - term: t statistic
    plain: >
      The estimate divided by its standard error. Values above about 2 are usually read as
      unlikely to be pure noise.

quiz:
  - kind: verbatim
    q: "The headline is +2.75. What does it count?"
    options:
      - "New Jersey restaurants hired 2.75 more people each."
      - "New Jersey's change in average full-time-equivalent staff per restaurant was 2.75 larger than Pennsylvania's change."
      - "Employment in New Jersey rose 2.75%."
    answer: 1
    why: It is a difference between two changes, in full-time-equivalent workers per restaurant.
  - kind: gist
    q: "Where does most of the 2.75 come from?"
    options:
      - "New Jersey restaurants adding staff."
      - "Pennsylvania restaurants losing staff over the same months."
    answer: 1
    why: New Jersey rose by about 0.6; Pennsylvania fell by about 2.2.
  - kind: transfer
    q: "A New Jersey restaurant closed for renovations during the second survey. How is it counted?"
    options: ["As zero workers", "It is left out of the second-round average", "As its first-round staff"]
    answer: 1
    why: Only permanently closed restaurants count as zero; the four temporary closures are treated as missing.

changelog:
  - {date: "1994-09-01", change: "Published in the American Economic Review 84(4)."}
  - {date: "2026-10-07", change: "Metric card written from the public data and the authors' check.sas; five discrepancies recorded."}

sources:
  - {id: S1, path: "examples/minimum-wage/data/njmin-aer.pdf", note: "Card & Krueger (1994), AER 84(4): intro, section I, Table 1, section III.A, Tables 3–4 notes"}
  - {id: S2, path: "examples/minimum-wage/data/check.sas", lines: "54-59, 128-139, 150-163"}
  - {id: S3, path: "examples/minimum-wage/data/codebook", note: "Column definitions; STATUS2 codes"}
  - {id: S4, path: "examples/minimum-wage/data/read.me", note: "Says the files support Myth and Measurement, chapter 2"}
  - {id: S5, path: "examples/minimum-wage/analysis.py", lines: "30-103", note: "Outputs in examples/minimum-wage/results/"}
  - {id: S6, path: "examples/minimum-wage/data/public.dat", note: "410 rows; downloaded from davidcard.berkeley.edu/data_sets.html"}

Estimand

Population	Burger King, KFC, Wendy's and Roy Rogers restaurants in New Jersey and eastern Pennsylvania in the 1992 sample frame that responded in wave 1 and reported employment in the given wave.
Exposure / contrast	Being in New Jersey when its minimum wage rose from $4.25 to $5.05 on 1992-04-01.
Time zero	1992-04-01.
Variable	FTE employment per store = full-time + managers + 0.5 × part-time, in wave 2 (Nov 5 – Dec 31, 1992) minus wave 1 (Feb 15 – Mar 4, 1992).
Intercurrent events	Permanent closures count as zero employment; temporary closures (4) are missing in wave 2.
Summary measure	Difference in state mean changes (unbalanced means), FTE workers per store; interval from the published t statistic.
Target	ATT: Effect among those who actually did it (not everyone)

S1S2S5

Rules as implemented

Stage	Exact rule	Source
Eligibility 1 410 remain	public.dat: 410 stores, STATE = 1 (NJ, 331) or 0 (PA, 79).	S1S3
Eligibility 2	Wave-1 mean over stores with non-missing fte1 (PA 77, NJ 321); wave-2 mean over stores with non-missing fte2 (PA 77, NJ 319). "All available observations" (Table 3 rows 1–2).	S1S5
Eligibility 3	STATUS2 = 3 (6 stores): wave-2 employment fields are already 0 in public.dat. STATUS2 in (2, 4, 5) (4 stores, all NJ): wave-2 fields missing, so excluded from the wave-2 mean.	S1S3S6
Time zero	1992-04-01 (paper, introduction and section I). Pennsylvania stayed at the federal $4.25.	S1
First survey round [-46, -27) days	Wave 1: February 15 – March 4, 1992 (Table 1).	S1
Second survey round [218, 275) days	Wave 2: November 5 – December 31, 1992 (Table 1); DATE2 in the file spans 110592–123192.	S1S6
Staff before value	EMPTOT = EMPPT*.5 + EMPFT + NMGRS (check.sas line 54).	S2S1
Staff after value	EMPTOT2 = EMPPT2*.5 + EMPFT2 + NMGRS2 (check.sas line 55).	S2
Aggregate	PROC MEANS of EMPTOT and EMPTOT2 BY NJ (check.sas lines 128–131, 'PART OF TABLE 3').	S2S5
Comparison	(NJ wave 2 − NJ wave 1) − (PA wave 2 − PA wave 1), FTE per store (Table 3, row 3, column iii).	S1S5
Uncertainty	SE = 2.76 / 2.03 ≈ 1.36 (paper, section III.A); interval = 2.754 ± 1.96 × 1.36. Treating the four state-by-round means as independent gives SE 1.80 (interval −0.78 to 6.28). The balanced-sample change (row 4) gives SE 1.34.	S1S5

Numbers per group

Group	Definition	Value	n
Pennsylvania, Feb–Mar	STATE = 0, non-missing EMPTOT (n = 77).	23.3312	77
Pennsylvania, Nov–Dec	STATE = 0, non-missing EMPTOT2 (n = 77).	21.1656	77
New Jersey, Feb–Mar	STATE = 1, non-missing EMPTOT (n = 321).	20.4394	321
New Jersey, Nov–Dec	STATE = 1, non-missing EMPTOT2 (n = 319).	21.0274	319

Check: (21.0274 − 20.4394) − (21.1656 − 23.3312) = 2.754 (headline 2.75).

Parameters

Name	Value	Meaning
`old_minimum`	$4.25	Federal and NJ rate before 1992-04-01
`new_minimum`	$5.05	NJ rate from 1992-04-01
`part_time_weight`	0.5	Each part-time worker counts as half
`temporarily_closed_status`	2, 4, 5	STATUS2 codes treated as missing in wave 2

Data lineage

public.dat
One row per store, both survey waves
Grain: store
check.sas
Authors' program; computes FTE and the per-state means of Table 3
Grain: store → state × wave
examples/minimum-wage/analysis.py
Python reproduction of Table 3 rows 1–5 and Table 4 column (i)
Grain: state × wave

Code that defines it

FTE employment in each wave (authors' SAS)examples/minimum-wage/data/check.sas:54-59, 128-139, 150-163

EMPTOT=EMPPT*.5 + EMPFT + NMGRS;
EMPTOT2=EMPPT2*.5 + EMPFT2 + NMGRS2;
DEMP=EMPTOT2-EMPTOT;;

Difference-in-differences and the independent-samples standard error (reproduction)examples/minimum-wage/analysis.py:30-103

did = (lv["NJ_post"]["value"] - lv["NJ_pre"]["value"]) - (lv["PA_post"]["value"] - lv["PA_pre"]["value"])
se_indep = np.sqrt(sum(r["se"] ** 2 for r in rows))

Rebuild it

python examples/fetch_data.py minimum-wage (downloads and checks the data and paper).
python examples/minimum-wage/analysis.py (writes results/table3.csv and results/checks.csv).

Receipts:

2026-10-07: PA 23.33 → 21.17, NJ 20.44 → 21.03; DiD 2.754. Row 4 2.750 (SE 1.342), row 5 2.509.
2026-10-07: Table 4 column (i) 2.326 (SE 1.192), n = 357, mean change −0.238, SD 8.825; the paper prints 2.33 (1.19), −0.237, 8.825.

Tests: examples/minimum-wage/results/checks.csv compares each reproduced figure with the paper.

Sources

S1 examples/minimum-wage/data/njmin-aer.pdf · Card & Krueger (1994), AER 84(4): intro, section I, Table 1, section III.A, Tables 3–4 notes
S2 examples/minimum-wage/data/check.sas lines 54-59, 128-139, 150-163
S3 examples/minimum-wage/data/codebook · Column definitions; STATUS2 codes
S4 examples/minimum-wage/data/read.me · Says the files support Myth and Measurement, chapter 2
S5 examples/minimum-wage/analysis.py lines 30-103 · Outputs in examples/minimum-wage/results/
S6 examples/minimum-wage/data/public.dat · 410 rows; downloaded from davidcard.berkeley.edu/data_sets.html

Change log

1994-09-01 Published in the American Economic Review 84(4).
2026-10-07 Metric card written from the public data and the authors' check.sas; five discrepancies recorded.

Card quality checks: 0 errors · 0 warnings

info recompute estimate reproduced from cells: (21.0274 − 20.4394) − (21.1656 − 23.3312) = 2.754

Glossary

95% range (95% interval, confidence interval): The values that fit the data well, given random noise. It does not cover the assumption that both states would have moved alike.
difference-in-differences (DiD): Compare one group's change with another group's change over the same time. The difference between the two changes is the part a shared trend does not explain.
full-time equivalent (FTE, full-time-equivalent): A staff count where each part-time worker counts as half a full-time worker. Managers count as full-time.
minimum wage: The lowest hourly pay an employer may legally offer.
t statistic: The estimate divided by its standard error. Values above about 2 are usually read as unlikely to be pure noise.

When New Jersey raised its minimum wage in April 1992, did fast-food restaurants there cut jobs compared with similar restaurants just across the border in Pennsylvania, where the wage did not change?

✓ Safe to say

✗ Not this

! Before you repeat it

How the number is made

Follow one restaurant: A Wendy's in Camden County

Start with restaurants, then apply each rule

Time zero: wage rise

Staff after

Average across people

Subtract the usual change

Show the range of plausible values