Literature Coverage

The mergers economists have studied differ from the universe of U.S. mergers. They are larger, concentrated in particular industries, and clustered in certain years. But that distribution lines up more closely with the mergers that drew antitrust scrutiny (HSR filings and second requests), which arguably are the more policy-relevant comparison. The charts below let you compare retrospectives against each of those benchmarks.

How representative is the merger-retrospective literature? Each chart compares the share of retrospectives in a category against the share of merger activity in that same category, for a benchmark you can choose. A bar pointing one way means a category is over-represented in the literature relative to that benchmark; the other way means it is under-represented.

Use the selector in each chart to choose which measure of merger activity to compare against:

• All mergers (universe): the full population of mergers.
• HSR filings: mergers large enough to require Hart–Scott–Rodino notification.
• Second Requests: mergers that drew heightened antitrust scrutiny.

navy = "#1c3d5a"
gold = "#b0843a"

By industry

industryRaw = FileAttachment("Descs_mergers_industry.csv").csv({typed: true})

viewof indBase = Inputs.select(
  new Map([
    ["All mergers (universe)", "x_Universe"],
    ["HSR filings", "x_HSR"],
    ["Second Requests", "x_SecondReq"]
  ]),
  {label: "Compare retrospectives to:", value: "x_Universe"}
)

industryData = industryRaw
  .map(d => ({
    industry: d.industry,
    gap: d.x_MetaData - d[indBase],
    retro: d.x_MetaData,
    activity: d[indBase]
  }))
  .sort((a, b) => d3.descending(a.gap, b.gap))

Plot.plot({
  marginLeft: 200,
  marginRight: 20,
  marginTop: 30,
  height: 420,
  x: {
    label: "share of retrospectives − share of merger activity",
    tickFormat: "+.0%",
    grid: true
  },
  y: {label: null, domain: industryData.map(d => d.industry)},
  marks: [
    Plot.barX(industryData, {
      y: "industry",
      x: "gap",
      fill: d => d.gap >= 0 ? navy : gold,
      channels: {
        "Industry": d => d.industry,
        "Retrospectives": d => (d.retro * 100).toFixed(1) + "%",
        "Merger activity": d => (d.activity * 100).toFixed(1) + "%",
        "Difference": d => (d.gap >= 0 ? "+" : "") + (d.gap * 100).toFixed(1) + "%"
      },
      tip: {format: {x: false, x1: false, x2: false, y: false, fill: false}}
    }),
    Plot.ruleX([0]),
    Plot.text(["← Under-studied"], {frameAnchor: "top-left", dx: 4, dy: -18, fill: "#5f6b76", fontSize: 11, fontStyle: "italic"}),
    Plot.text(["Over-studied →"], {frameAnchor: "top-right", dx: -4, dy: -18, fill: "#5f6b76", fontSize: 11, fontStyle: "italic"})
  ]
})

Over time

yearRaw = FileAttachment("Descs_mergers_year.csv").csv({typed: true})

yearRows = yearRaw.filter(d => d.fy != null)

yearLines = yearRows.flatMap(d => [
  d.universe_rel != null ? {fy: d.fy, series: "Universe", value: d.universe_rel, raw: d.universe} : null,
  d.hsr_rel != null ? {fy: d.fy, series: "HSR filings", value: d.hsr_rel, raw: d.hsr} : null,
  d.second_requests_rel != null ? {fy: d.fy, series: "Second Requests", value: d.second_requests_rel, raw: d.second_requests} : null
]).filter(d => d != null)

yearBarMax = d3.max(yearRows, d => d.sample) || 1
yearLineMax = d3.max(yearLines, d => d.value) || 100
yearScale = yearLineMax / yearBarMax

Plot.plot({
  height: 380,
  marginLeft: 80,
  marginRight: 70,
  x: {label: "Fiscal year", tickFormat: "d"},
  y: {grid: true, domain: [0, yearLineMax * 1.05]},
  color: {
    domain: ["Universe", "HSR filings", "Second Requests"],
    range: ["#1c3d5a", "#b0843a", "#7b2d3b"],
    legend: true
  },
  marks: [
    Plot.axisY({
      anchor: "left",
      label: "↑ Normalized merger activity (1990 = 100)",
      labelAnchor: "center"
    }),
    Plot.axisY({
      anchor: "right",
      label: "↑ Retrospectives in sample",
      labelAnchor: "center",
      ticks: d3.range(0, yearBarMax + 1).map(b => b * yearScale),
      tickFormat: t => Math.round(t / yearScale)
    }),
    Plot.ruleY([0]),
    Plot.rectY(yearRows, {
      x1: d => d.fy - 0.42,
      x2: d => d.fy + 0.42,
      y: d => d.sample * yearScale,
      fill: "#9ca3af",
      stroke: "#6b7280",
      channels: {
        "Year": d => String(d.fy),
        "Retrospectives in sample": d => d.sample
      },
      tip: {format: {x1: false, x2: false, y: false, fill: false}}
    }),
    Plot.lineY(yearLines, {x: "fy", y: "value", stroke: "series", strokeWidth: 2.2}),
    Plot.dot(yearLines, {
      x: "fy", y: "value", fill: "series", r: 3.5,
      stroke: "white", strokeWidth: 1,
      channels: {
        "Year": d => String(d.fy),
        "Series": "series",
        "Index (1990 = 100)": d => d.value.toFixed(1),
        "Count": d => d.raw != null ? d3.format(",")(d.raw) : "n/a"
      },
      tip: {format: {x: false, y: false, fill: false}}
    })
  ]
})

By deal value

valueRaw = FileAttachment("Descs_mergers_value_hist.csv").csv({typed: true})

bucketOrder = ["<50M", "50-100M", "150-200M", "200-300M", "300-500M", "500M-1B", ">1B"]

viewof valBase = Inputs.select(
  new Map([
    ["All mergers (universe)", "frac_eikon"],
    ["HSR filings", "hsr_number_frac"],
    ["Second Requests", "second_tot_frac"]
  ]),
  {label: "Compare retrospectives to:", value: "frac_eikon"}
)

valueData = valueRaw
  .map(d => ({
    bucket: d.bucket,
    gap: d.frac_sample - d[valBase],
    retro: d.frac_sample,
    activity: d[valBase]
  }))
  .sort((a, b) => d3.ascending(bucketOrder.indexOf(a.bucket), bucketOrder.indexOf(b.bucket)))

Plot.plot({
  height: 340,
  marginLeft: 80,
  marginTop: 30,
  x: {label: "share of retrospectives − share of merger activity", tickFormat: "+.0%", grid: true},
  y: {label: "Deal value", domain: bucketOrder},
  marks: [
    Plot.barX(valueData, {
      y: "bucket",
      x: "gap",
      fill: d => d.gap >= 0 ? navy : gold,
      channels: {
        "Deal value": d => d.bucket,
        "Retrospectives": d => (d.retro * 100).toFixed(1) + "%",
        "Merger activity": d => (d.activity * 100).toFixed(1) + "%",
        "Difference": d => (d.gap >= 0 ? "+" : "") + (d.gap * 100).toFixed(1) + "%"
      },
      tip: {format: {x: false, x1: false, x2: false, y: false, fill: false}}
    }),
    Plot.ruleX([0]),
    Plot.text(["← Under-studied"], {frameAnchor: "top-left", dx: 4, dy: -18, fill: "#5f6b76", fontSize: 11, fontStyle: "italic"}),
    Plot.text(["Over-studied →"], {frameAnchor: "top-right", dx: -4, dy: -18, fill: "#5f6b76", fontSize: 11, fontStyle: "italic"})
  ]
})

Gaps in the literature

Reading the patterns together, the gaps the paper highlights are:

1. High-technology and other business-to-business sectors, where price data are typically not publicly available and retrospective study has been limited.
2. Smaller deals, which make up most merger activity but are heavily under-represented relative to mergers above $1B.
3. More recent mergers, since the literature drops off sharply after the mid-2000s, partly because retrospectives require several post-merger years of data, and partly because recent work has shifted toward large-class studies of entire industries.

If your work covers one of these areas, submit it to the repository.