The graduate squeeze: young graduates, unemployment and AI exposure in Africa

Evidence from Ghana, Nigeria, Kenya, Egypt, Senegal and South Africa

Emmanuel Edudzie, October 2026. Analysis code written with substantial help from Claude (Anthropic), under the author's direction.

The question. The first note in this series found that few young workers hold jobs exposed to generative AI, and that the exposure sits in office work. The second found that students are a large share of early AI users. This note asks about the people in between: young graduates. Do they find work, and is the work they find the kind that generative AI could change?

Graduates here means people with advanced education in the ILO's aggregate levels, which is tertiary education: ISCED 2011 levels 5–8 (short-cycle tertiary, bachelor's, master's, doctoral). Section 1 checks this against the detailed ISCED levels in the data.

Data sources

File	What it is
UNE_DEAP_SEX_AGE_EDU_RT_A.csv	ILOSTAT: unemployment rate by sex, age and education (%), annual
EMP_TEMP_SEX_AGE_EDU_NB_A.csv	ILOSTAT: employment by sex, age and education (thousands), annual
EMP_TEMP_SEX_OCU_EDU_NB_A.csv	ILOSTAT: employment by sex, occupation and education (thousands), annual

All three ILOSTAT files were downloaded on 30 October 2026 for Egypt, Ghana, Kenya, Nigeria, Senegal and South Africa, 2015 onwards.

0. Set-up

from decimal import ROUND_HALF_UP, Decimal
from pathlib import Path

import matplotlib.pyplot as plt
import numpy as np
import pandas as pd

ROOT = Path.cwd()
RAW, CLEAN, FIG = ROOT / "data" / "raw", ROOT / "data" / "clean", ROOT / "outputs" / "figures"
for folder in (CLEAN, FIG):
    folder.mkdir(parents=True, exist_ok=True)
pd.set_option("display.width", 160)

SURFACE, INK, INK_2, GRID = "#fcfcfb", "#0b0b0b", "#52514e", "#e4e3df"
BLUE, GREY, REFERENCE = "#2a78d6", "#c3c2b7", "#8a8984"
plt.rcParams.update({
    "figure.facecolor": SURFACE, "axes.facecolor": SURFACE, "savefig.facecolor": SURFACE,
    "axes.edgecolor": GRID, "axes.labelcolor": INK_2, "xtick.color": INK_2, "ytick.color": INK,
    "text.color": INK, "font.size": 10, "axes.spines.top": False, "axes.spines.right": False,
})


def frame(fig, title, subtitle, source):
    """Title and subtitle at the top left, source line at the bottom left."""
    fig.text(0.01, 0.985, title, fontsize=12, fontweight="bold", color=INK, va="top")
    fig.text(0.01, 0.925, subtitle, fontsize=9.5, color=INK_2, va="top")
    fig.text(0.01, 0.01, source, fontsize=7.5, color=INK_2, va="bottom")


def half_up(x, places=1):
    """Round halves up (12.25 -> 12.3). Python's round() can go either way."""
    return float(Decimal(str(round(float(x), 6))).quantize(Decimal(1).scaleb(-places), rounding=ROUND_HALF_UP))


# The latest survey year in each file for every country (section 1 checks the sources)
SELECTED = {"Ghana": 2025, "Nigeria": 2025, "Kenya": 2021, "Egypt": 2024, "Senegal": 2025, "South Africa": 2025}
COUNTRIES = list(SELECTED)
EDU_ORDER = ["Less than basic", "Basic", "Intermediate", "Advanced"]
GROUP_NAMES = {
    "1": "Managers", "2": "Professionals", "3": "Technicians and associate professionals",
    "4": "Clerical support workers", "5": "Service and sales workers",
    "6": "Skilled agricultural, forestry and fishery workers", "7": "Craft and related trades workers",
    "8": "Plant and machine operators, and assemblers", "9": "Elementary occupations",
}


def load(name):
    """Read an ILOSTAT file, keep the chosen country-years and the aggregate education levels."""
    d = pd.read_csv(RAW / f"{name}.csv")
    d.columns = ["country", "source", "indicator", "sex", "classif1", "edu", "time", "value", "status",
                 "note_classif", "note_indicator", "note_source"]
    keep = np.array([SELECTED.get(c) == t for c, t in zip(d["country"], d["time"])])
    d = d[keep & d["edu"].str.startswith("Education (Aggregate levels)").to_numpy()].copy()
    d["edu"] = d["edu"].str.split(": ").str[1]
    return d

1. Load the data and check it

1a. Sources and years

Every country's latest year comes from a labour force survey, except Kenya, whose latest is the 2021 Continuous Household Survey.

une = load("UNE_DEAP_SEX_AGE_EDU_RT_A")
emp_age = load("EMP_TEMP_SEX_AGE_EDU_NB_A")
emp_occ = load("EMP_TEMP_SEX_OCU_EDU_NB_A")
pd.concat({"unemployment": une.groupby(["country", "time"])["source"].first(),
           "employment by age": emp_age.groupby(["country", "time"])["source"].first(),
           "employment by occupation": emp_occ.groupby(["country", "time"])["source"].first()}, axis=1)

		unemployment	employment by age	employment by occupation
country	time
Egypt	2024	LFS - Labour Force Sample Survey	LFS - Labour Force Sample Survey	LFS - Labour Force Sample Survey
Ghana	2025	LFS - Labour Force Survey	LFS - Labour Force Survey	LFS - Labour Force Survey
Kenya	2021	HS - Continuous household survey	HS - Continuous household survey	HS - Continuous household survey
Nigeria	2025	LFS - Unemployment, Under-employment Watch	LFS - Unemployment, Under-employment Watch	LFS - Unemployment, Under-employment Watch
Senegal	2025	LFS - National Employment Survey	LFS - National Employment Survey	LFS - National Employment Survey
South Africa	2025	LFS - Quarterly Labour Force Survey	LFS - Quarterly Labour Force Survey	LFS - Quarterly Labour Force Survey

1b. "Advanced" means tertiary education

The detailed ISCED 2011 levels 5–8 should add up to the "Advanced" aggregate. We check it on total employment aged 15+ in each country that reports ISCED 2011.

raw_age = pd.read_csv(RAW / "EMP_TEMP_SEX_AGE_EDU_NB_A.csv")
raw_age = raw_age[np.array([SELECTED.get(c) == t for c, t in zip(raw_age["ref_area.label"], raw_age["time"])])
                  & (raw_age["sex.label"] == "Total") & (raw_age["classif1.label"] == "Age (Youth, adults): 15+")]
rows = []
for country, d in raw_age.groupby("ref_area.label"):
    v = d.set_index("classif2.label")["obs_value"]
    isced = v[v.index.str.startswith("Education (ISCED-11)")]
    if len(isced):
        rows.append({"country": country, "ISCED 5-8 (000s)": isced[isced.index.str.contains(r": [5-8]\.")].sum(),
                     "Advanced (000s)": v["Education (Aggregate levels): Advanced"]})
pd.DataFrame(rows).round(1)

	country	ISCED 5-8 (000s)	Advanced (000s)
0	Egypt	6107.6	6107.6
1	Ghana	1297.4	1297.4
2	Kenya	1045.8	1045.8
3	Nigeria	7919.2	7919.2
4	Senegal	279.1	279.1
5	South Africa	3426.3	3426.3

1c. Cells ILOSTAT flags as unreliable

Small samples are flagged. We list the flagged cells that touch the figures used below.

flag_une = une[(une["status"] == "Unreliable") & une["classif1"].isin(
    ["Age (Youth, adults): 15-24", "Age (10-year bands): 25-34"]) & une["edu"].isin(["Advanced", "Total"])]
flag_occ = emp_occ[(emp_occ["status"] == "Unreliable") & (emp_occ["edu"] == "Advanced")]
display(flag_une[["country", "sex", "classif1", "edu", "value"]])
flag_occ[["country", "sex", "classif1", "value"]]

	country	sex	classif1	edu	value
10572	Kenya	Male	Age (Youth, adults): 15-24	Advanced	18.737
16401	Senegal	Male	Age (Youth, adults): 15-24	Advanced	21.998
16422	Senegal	Female	Age (Youth, adults): 15-24	Advanced	36.777

	country	sex	classif1	value
336	Egypt	Female	Occupation (ISCO-08): 6. Skilled agricultural,...	4.338
6496	Ghana	Total	Occupation (Skill level): X - Not elsewhere cl...	4.601
6523	Ghana	Male	Occupation (Skill level): X - Not elsewhere cl...	4.601
6545	Ghana	Female	Occupation (Skill level): Skill level 1 ~ low	4.609
6550	Ghana	Female	Occupation (Skill level): X - Not elsewhere cl...	NaN
6762	Ghana	Total	Occupation (ISCO-08): 0. Armed forces occupations	4.601
6821	Ghana	Male	Occupation (ISCO-08): 0. Armed forces occupations	4.601
6870	Ghana	Female	Occupation (ISCO-08): 8. Plant and machine ope...	2.344
6875	Ghana	Female	Occupation (ISCO-08): 9. Elementary occupations	4.609
6880	Ghana	Female	Occupation (ISCO-08): 0. Armed forces occupations	NaN
11980	Kenya	Total	Occupation (Skill level): X - Not elsewhere cl...	NaN
12010	Kenya	Male	Occupation (Skill level): X - Not elsewhere cl...	NaN
12040	Kenya	Female	Occupation (Skill level): X - Not elsewhere cl...	NaN
12261	Kenya	Total	Occupation (ISCO-08): X. Not elsewhere classified	NaN
12308	Kenya	Male	Occupation (ISCO-08): 7. Craft and related tra...	8.185
12326	Kenya	Male	Occupation (ISCO-08): X. Not elsewhere classified	NaN
12373	Kenya	Female	Occupation (ISCO-08): 7. Craft and related tra...	3.993
12391	Kenya	Female	Occupation (ISCO-08): X. Not elsewhere classified	NaN
13791	Nigeria	Total	Occupation (ISCO-08): 0. Armed forces occupations	33.415
13848	Nigeria	Male	Occupation (ISCO-08): 0. Armed forces occupations	19.481
13897	Nigeria	Female	Occupation (ISCO-08): 8. Plant and machine ope...	28.539
13905	Nigeria	Female	Occupation (ISCO-08): 0. Armed forces occupations	NaN
15091	Senegal	Female	Occupation (Skill level): Skill level 1 ~ low	2.050
15295	Senegal	Total	Occupation (ISCO-08): 6. Skilled agricultural,...	2.608
15300	Senegal	Total	Occupation (ISCO-08): 7. Craft and related tra...	3.267
15306	Senegal	Total	Occupation (ISCO-08): 8. Plant and machine ope...	3.137
15316	Senegal	Total	Occupation (ISCO-08): 0. Armed forces occupations	4.638
15362	Senegal	Male	Occupation (ISCO-08): 6. Skilled agricultural,...	2.208
15367	Senegal	Male	Occupation (ISCO-08): 7. Craft and related tra...	2.460
15373	Senegal	Male	Occupation (ISCO-08): 8. Plant and machine ope...	2.968
15383	Senegal	Male	Occupation (ISCO-08): 0. Armed forces occupations	4.166
15401	Senegal	Female	Occupation (ISCO-08): 1. Managers	2.958
15424	Senegal	Female	Occupation (ISCO-08): 5. Service and sales wor...	5.363
15429	Senegal	Female	Occupation (ISCO-08): 6. Skilled agricultural,...	NaN
15434	Senegal	Female	Occupation (ISCO-08): 7. Craft and related tra...	NaN
15440	Senegal	Female	Occupation (ISCO-08): 8. Plant and machine ope...	NaN
15445	Senegal	Female	Occupation (ISCO-08): 9. Elementary occupations	2.050
15450	Senegal	Female	Occupation (ISCO-08): 0. Armed forces occupations	NaN
20727	South Africa	Total	Occupation (ISCO-88): 6. Skilled agricultural ...	5.601
20750	South Africa	Total	Occupation (ISCO-88): 0. Armed forces	3.942
20797	South Africa	Male	Occupation (ISCO-88): 6. Skilled agricultural ...	4.760
20820	South Africa	Male	Occupation (ISCO-88): 0. Armed forces	3.942
20867	South Africa	Female	Occupation (ISCO-88): 6. Skilled agricultural ...	NaN
20890	South Africa	Female	Occupation (ISCO-88): 0. Armed forces	NaN

None of the flagged cells is a headline total. The figures for graduate men and women aged 15–24 in Kenya and Senegal rest on flagged cells, so the sex split below uses ages 25–34.

2. Are young graduates more likely to be unemployed?

2a. Unemployment rate by education, ages 15–24 and 25–34

Few 15–24-year-olds have finished tertiary education, so ages 25–34 are the main measure for young graduates; 15–24 is shown alongside.

AGES = {"Age (Youth, adults): 15-24": "15–24", "Age (10-year bands): 25-34": "25–34"}
u = une[une["classif1"].isin(AGES)].assign(age=lambda d: d["classif1"].map(AGES))
rates = (u[u["sex"] == "Total"].pivot_table(index=["country", "age"], columns="edu", values="value")
         [EDU_ORDER + ["Total"]].reindex(COUNTRIES, level=0))
rates.to_csv(CLEAN / "unemployment_by_education.csv")
rates.round(1)

	edu	Less than basic	Basic	Intermediate	Advanced	Total
country	age
Ghana	15–24	3.5	4.0	7.4	13.2	5.9
	25–34	1.7	2.5	4.9	10.7	4.4
Nigeria	15–24	3.8	4.6	7.2	18.5	7.0
	25–34	2.8	3.7	4.5	16.9	5.9
Kenya	15–24	9.6	8.2	14.6	22.8	12.3
	25–34	4.3	5.0	10.0	11.7	8.1
Egypt	15–24	5.6	6.0	13.9	42.8	15.3
	25–34	2.6	3.0	6.0	17.7	8.5
Senegal	15–24	2.3	10.2	17.6	29.8	5.0
	25–34	3.2	9.4	21.3	22.0	7.8
South Africa	15–24	46.3	52.4	56.2	35.8	53.2
	25–34	34.2	44.2	37.7	21.4	37.3

gap = rates[["Advanced", "Total"]].copy()
gap["graduates ÷ all"] = gap["Advanced"] / gap["Total"]
gap.round(1)

	edu	Advanced	Total	graduates ÷ all
country	age
Ghana	15–24	13.2	5.9	2.2
	25–34	10.7	4.4	2.5
Nigeria	15–24	18.5	7.0	2.7
	25–34	16.9	5.9	2.9
Kenya	15–24	22.8	12.3	1.9
	25–34	11.7	8.1	1.5
Egypt	15–24	42.8	15.3	2.8
	25–34	17.7	8.5	2.1
Senegal	15–24	29.8	5.0	5.9
	25–34	22.0	7.8	2.8
South Africa	15–24	35.8	53.2	0.7
	25–34	21.4	37.3	0.6

Reading this. In five of the six countries, graduates are more likely than their peers to be unemployed, at both ages. South Africa is the exception: unemployment is very high for everyone, and lowest for graduates. A common explanation for the pattern elsewhere is that where most work is informal and there are no unemployment benefits, people with little education cannot afford to stay unemployed, while graduates can wait longer for the formal jobs they trained for.

2b. Graduate unemployment by sex, ages 25–34

by_sex = (u[(u["age"] == "25–34") & (u["edu"] == "Advanced")]
          .pivot_table(index="country", columns="sex", values="value")[["Female", "Male", "Total"]]
          .reindex(COUNTRIES))
by_sex.to_csv(CLEAN / "graduate_unemployment_by_sex_25_34.csv")
by_sex.round(1)

sex	Female	Male	Total
country
Ghana	11.0	10.6	10.7
Nigeria	16.9	17.0	16.9
Kenya	12.6	10.8	11.7
Egypt	31.0	10.6	17.7
Senegal	25.2	20.2	22.0
South Africa	24.7	16.9	21.4

2c. Chart: unemployment of graduates and of all young people

fig, axes = plt.subplots(1, 2, figsize=(8, 4.2), sharey=True)
fig.subplots_adjust(left=0.15, right=0.97, top=0.74, bottom=0.14, wspace=0.14)
order = rates.xs("25–34", level="age")["Advanced"].sort_values().index.tolist()   # bottom to top
for ax, age in zip(axes, ["15–24", "25–34"]):
    r = rates.xs(age, level="age").loc[order]
    for y, (country, row) in enumerate(r.iterrows()):
        ax.plot([row["Total"], row["Advanced"]], [y, y], color=GRID, linewidth=3, zorder=1)
        ax.scatter(row["Total"], y, s=60, color=GREY, zorder=2, label="All in the age group" if y == 0 else None)
        ax.scatter(row["Advanced"], y, s=60, color=BLUE, zorder=3, label="Graduates" if y == 0 else None)
        label = f"{half_up(row['Advanced'], 0):.0f}"
        if row["Advanced"] >= row["Total"]:     # graduate dot on the right: label to its right
            ax.text(row["Advanced"] + 2.2, y, label, va="center", ha="left", fontsize=8.5, color=BLUE, fontweight="bold")
        else:                                   # South Africa: graduate dot on the left
            ax.text(row["Advanced"] - 2.2, y, label, va="center", ha="right", fontsize=8.5, color=BLUE, fontweight="bold")
    ax.set_title(f"Aged {age}", fontsize=9.5, loc="left", fontweight="bold", color=INK)
    ax.set_xlim(0, 62)
    ax.set_xticks([0, 20, 40, 60], ["0", "20", "40", "60%"])
    ax.grid(axis="x", color=GRID, linewidth=0.8)
    ax.set_axisbelow(True)
    ax.spines[["left", "bottom"]].set_visible(False)
    ax.tick_params(axis="both", length=0)
    ax.set_yticks(range(len(order)), order)
axes[0].legend(frameon=False, loc="lower left", bbox_to_anchor=(0.0, 1.08), ncol=2, fontsize=8.5,
               handletextpad=0.2, columnspacing=1.2)
frame(fig, "Young graduates are more likely to be unemployed than their peers, except in South Africa",
      "Unemployment rate (%): people with tertiary education against everyone in the age group",
      "Source: ILOSTAT, unemployment rate by sex, age and education (Ghana 2025, Nigeria 2025, Kenya 2021, Egypt 2024,\n"
      "Senegal 2025, South Africa 2025). Graduates = advanced (tertiary) education, ISCED 2011 levels 5–8.")
fig.savefig(FIG / "graduate_unemployment.png", dpi=200)
plt.show()

3. How many young workers are graduates?

Share of employed people with tertiary education, by age.

AGE3 = {"Age (Youth, adults): 15-24": "15–24", "Age (10-year bands): 25-34": "25–34", "Age (Youth, adults): 25+": "25+"}
e = emp_age[(emp_age["sex"] == "Total") & emp_age["classif1"].isin(AGE3)].assign(age=lambda d: d["classif1"].map(AGE3))
e = e.pivot_table(index=["country", "age"], columns="edu", values="value")
grad_share = (e["Advanced"] / e["Total"] * 100).unstack("age").reindex(COUNTRIES)
grad_share.to_csv(CLEAN / "graduate_share_of_employment.csv")
grad_share.round(1)

age	15–24	25+	25–34
country
Ghana	5.6	10.0	14.1
Nigeria	4.1	9.8	11.7
Kenya	3.4	5.7	7.9
Egypt	10.3	20.1	25.6
Senegal	1.3	6.7	8.7
South Africa	8.6	19.6	18.1

4. Where do employed graduates work?

ILOSTAT publishes employment by occupation and education for all ages together, not by age, so this section covers all employed graduates. It uses the nine major occupation groups and the exposure tiers from the first note:

Tier	Occupation groups
High	4 Clerical support
Mixed	2 Professionals, 3 Technicians (finance, ICT and media highly exposed; teaching and health not)
Partial	5 Service and sales
Low	1 Managers, 6 Farming, 7 Crafts, 8 Machine operators, 9 Elementary work

Senegal is excluded here: 65% of its workers' occupations are unclassified. South Africa reports the older ISCO-88 groups, as in the first note. Unclassified workers and the armed forces are left out of the denominators.

o = emp_occ[emp_occ["country"] != "Senegal"].copy()
o["scheme"] = o["classif1"].str.extract(r"\((ISCO-08|ISCO-88)\)")[0]
o = o.dropna(subset=["scheme"])
preferred = o.groupby("country")["scheme"].agg(lambda s: "ISCO-08" if "ISCO-08" in set(s) else "ISCO-88")
o = o[o["scheme"] == o["country"].map(preferred)]
o["group"] = o["classif1"].str.split(": ").str[1].str.split(".").str[0]

total = o[(o["group"] == "Total") & (o["sex"] == "Total")].set_index(["country", "edu"])["value"]
unclassified = o[(o["group"] == "X") & (o["sex"] == "Total")].set_index(["country", "edu"])["value"]
print("Unclassified occupation, % of employed:")
display((unclassified / total * 100).unstack("edu").reindex(columns=EDU_ORDER + ["Total"]).round(1))

nine = o[o["group"].isin(list("123456789"))].copy()
nine["share"] = nine["value"] / nine.groupby(["country", "sex", "edu"])["value"].transform("sum") * 100
nine.to_csv(CLEAN / "employment_shares_by_group_and_education.csv", index=False)

TIERS = {"4": "High: clerical support", "2": "Mixed: professionals and technicians",
         "3": "Mixed: professionals and technicians", "5": "Partial: service and sales"}
TIER_ORDER = ["High: clerical support", "Mixed: professionals and technicians",
              "Partial: service and sales", "Low: farming, crafts, operators, elementary, managers"]
nine["tier"] = nine["group"].map(TIERS).fillna(TIER_ORDER[3])
tiers = nine.groupby(["country", "sex", "edu", "tier"])["share"].sum().unstack("tier")[TIER_ORDER]
tiers.round(2).to_csv(CLEAN / "employment_shares_by_tier_and_education.csv")
tiers.xs("Total", level="sex").round(1)

Unclassified occupation, % of employed:

edu	Less than basic	Basic	Intermediate	Advanced	Total
country
Egypt	14.5	12.8	6.0	2.6	8.4
Ghana	NaN	NaN	NaN	NaN	NaN
Kenya	2.0	1.2	0.9	NaN	1.1
Nigeria	1.6	2.5	3.1	3.1	2.8
South Africa	26.5	9.3	3.2	1.0	6.2

	tier	High: clerical support	Mixed: professionals and technicians	Partial: service and sales	Low: farming, crafts, operators, elementary, managers
country	edu
Egypt	Advanced	8.8	60.3	12.8	18.1
	Basic	0.6	2.1	19.6	77.6
	Intermediate	6.5	14.2	20.7	58.7
	Less than basic	0.2	1.0	17.5	81.3
	Total	4.7	18.7	18.3	58.4
Ghana	Advanced	6.8	49.9	15.0	28.3
	Basic	0.4	1.2	27.7	70.7
	Intermediate	3.2	6.7	32.0	58.1
	Less than basic	0.0	0.4	16.3	83.3
	Level not stated	NaN	NaN	58.5	41.5
	Total	1.5	6.7	25.4	66.4
Kenya	Advanced	3.1	52.3	2.8	41.8
	Basic	0.1	2.1	6.2	91.6
	Intermediate	1.7	12.3	12.0	74.0
	Less than basic	NaN	2.6	3.7	93.7
	Level not stated	0.0	20.7	0.0	79.3
	Total	0.9	9.1	8.2	81.7
Nigeria	Advanced	4.6	32.5	27.2	35.6
	Basic	0.3	1.7	30.8	67.3
	Intermediate	0.7	6.4	33.8	59.1
	Less than basic	NaN	1.5	25.1	73.4
	Total	0.9	7.1	32.0	59.9
South Africa	Advanced	13.4	49.7	8.2	28.7
	Basic	3.6	2.3	16.2	77.8
	Intermediate	16.3	9.1	23.6	51.0
	Less than basic	0.6	1.7	8.0	89.7
	Level not stated	5.7	8.9	13.1	72.2
	Total	10.8	14.7	17.3	57.3

OCC_COUNTRIES = [c for c in COUNTRIES if c != "Senegal"]
t = tiers.xs("Total", level="sex")
office = pd.DataFrame({"Graduates": t.xs("Advanced", level="edu")[TIER_ORDER[:2]].sum(axis=1),
                       "All workers": t.xs("Total", level="edu")[TIER_ORDER[:2]].sum(axis=1)}).reindex(OCC_COUNTRIES)
# basic or less: recombine the two lowest levels by numbers employed, not by averaging shares
low_edu = nine[(nine["sex"] == "Total") & nine["edu"].isin(["Less than basic", "Basic"])]
low_edu = low_edu.assign(office=low_edu["tier"].isin(TIER_ORDER[:2]))
office["Basic or less"] = (low_edu[low_edu["office"]].groupby("country")["value"].sum()
                           / low_edu.groupby("country")["value"].sum() * 100)
office["graduates ÷ all"] = office["Graduates"] / office["All workers"]
office.round(1)

	Graduates	All workers	Basic or less	graduates ÷ all
country
Ghana	56.7	8.2	1.3	6.9
Nigeria	37.1	8.0	1.8	4.6
Kenya	55.4	10.1	2.3	5.5
Egypt	69.1	23.4	1.8	3.0
South Africa	63.0	25.5	5.5	2.5

Graduates are far more likely than other workers to hold clerical, professional and technical jobs, the tiers where generative AI exposure is concentrated. Part of the "mixed" tier is teaching and health, which are not exposed, so this is an upper bound on graduates' exposure.

4b. Chart: exposure tiers of graduates' jobs and of all jobs

TIER_STYLE = {TIER_ORDER[0]: (BLUE, 1.0), TIER_ORDER[1]: (BLUE, 0.62), TIER_ORDER[2]: (BLUE, 0.3), TIER_ORDER[3]: (GREY, 1.0)}
SHORT = {TIER_ORDER[0]: "High: clerical", TIER_ORDER[1]: "Mixed: professionals, technicians",
         TIER_ORDER[2]: "Partial: service, sales", TIER_ORDER[3]: "Low: other"}
fig, ax = plt.subplots(figsize=(8, 5.4))
fig.subplots_adjust(left=0.2, right=0.86, top=0.8, bottom=0.1)
labels, ypos, data = [], [], []
for i, country in enumerate(OCC_COUNTRIES[::-1]):
    for j, (edu, name) in enumerate([("Total", "All workers"), ("Advanced", "Graduates")]):
        y = i * 2.6 + j
        ypos.append(y)
        labels.append(f"{country}  {name}" if j == 1 else name)
        data.append(t.loc[(country, edu)])
data = pd.DataFrame(data)
left = np.zeros(len(data))
for tier in TIER_ORDER:
    colour, alpha = TIER_STYLE[tier]
    ax.barh(ypos, data[tier], left=left, color=colour, alpha=alpha, edgecolor=SURFACE, linewidth=1.2, height=0.8,
            label=SHORT[tier])
    left += data[tier].to_numpy()
for y, v in zip(ypos, data[TIER_ORDER[:2]].sum(axis=1)):
    ax.text(103, y, f"{half_up(v, 0):.0f}%", va="center", fontsize=9, fontweight="bold", color=BLUE)
ax.text(103, max(ypos) + 1.1, "Office-based\n(high + mixed)", fontsize=8, color=INK_2, va="bottom")
ax.set_yticks(ypos, [lab.replace("  ", "\n") if "Graduates" in lab else lab for lab in labels], fontsize=8.5)
ax.set_xlim(0, 100)
ax.set_xticks([0, 25, 50, 75, 100])
ax.tick_params(axis="both", length=0)
ax.spines[["left", "bottom"]].set_visible(False)
ax.legend(frameon=False, loc="lower left", bbox_to_anchor=(-0.22, 1.0), ncol=4, fontsize=8, handlelength=1.2,
          columnspacing=1.0)
frame(fig, "Graduates hold the office jobs where generative AI exposure is concentrated",
      "Share of employed people with a known occupation, by exposure tier of their occupation group (%)",
      "Source: ILOSTAT, employment by sex, occupation and education (Ghana 2025, Nigeria 2025, Kenya 2021, Egypt 2024,\n"
      "South Africa 2025); tiers from ILO Working Paper 140. All ages. South Africa uses ISCO-88 major groups.")
fig.savefig(FIG / "graduate_exposure_tiers.png", dpi=200)
plt.show()

5. How many graduates work below graduate level?

The ILO's standard (normative) measure links occupation groups to skill levels: groups 1–3 (managers, professionals, technicians) are skill levels 3–4, which usually call for tertiary education; groups 4–9 are skill levels 1–2. A graduate in groups 4–9 is counted as working below graduate level ("overeducated").

grads = nine[(nine["sex"] == "Total") & (nine["edu"] == "Advanced")].pivot_table(
    index="country", columns="group", values="share").reindex(OCC_COUNTRIES)
below = pd.DataFrame({
    "Clerical (group 4)": grads["4"],
    "Service and sales (group 5)": grads["5"],
    "Farming, crafts, operators, elementary (6–9)": grads[["6", "7", "8", "9"]].sum(axis=1),
})
below["Below graduate level, total"] = below.sum(axis=1)
below.to_csv(CLEAN / "graduates_below_graduate_level.csv")
below.round(1)

	Clerical (group 4)	Service and sales (group 5)	Farming, crafts, operators, elementary (6–9)	Below graduate level, total
country
Ghana	6.8	15.0	18.3	40.1
Nigeria	4.6	27.2	31.2	63.0
Kenya	3.1	2.8	18.0	23.9
Egypt	8.8	12.8	10.3	31.9
South Africa	13.4	8.2	7.8	29.4

by_sex_below = (nine[(nine["edu"] == "Advanced") & nine["group"].isin(list("456789"))]
                .groupby(["country", "sex"])["share"].sum().unstack("sex")[["Female", "Male"]].reindex(OCC_COUNTRIES))
by_sex_below.round(1)

sex	Female	Male
country
Ghana	42.3	38.5
Nigeria	65.1	61.8
Kenya	17.2	28.4
Egypt	16.3	38.3
South Africa	31.6	26.6

fig, ax = plt.subplots(figsize=(8, 3.9))
fig.subplots_adjust(left=0.16, right=0.95, top=0.72, bottom=0.12)
b = below.sort_values("Below graduate level, total")
parts = [("Clerical (group 4)", BLUE, 1.0, "Clerical (high exposure)"),
         ("Service and sales (group 5)", BLUE, 0.35, "Service and sales"),
         ("Farming, crafts, operators, elementary (6–9)", GREY, 1.0, "Farming, crafts, operators, elementary")]
left = np.zeros(len(b))
for col, colour, alpha, name in parts:
    ax.barh(b.index, b[col], left=left, color=colour, alpha=alpha, edgecolor=SURFACE, linewidth=1.2, height=0.65,
            label=name)
    left += b[col].to_numpy()
for y, v in enumerate(b["Below graduate level, total"]):
    ax.text(v + 1, y, f"{half_up(v, 0):.0f}%", va="center", fontsize=9.5, fontweight="bold", color=INK)
ax.set_xlim(0, 75)
ax.set_xticks([])
ax.spines[["left", "bottom"]].set_visible(False)
ax.tick_params(axis="y", length=0, labelsize=10)
ax.legend(frameon=False, loc="lower left", bbox_to_anchor=(-0.2, 1.0), ncol=3, fontsize=8, handlelength=1.2)
frame(fig, "In Nigeria, nearly two in three employed graduates work below graduate level",
      "Share of employed graduates in occupations that do not usually require tertiary education (%), by occupation",
      "Source: ILOSTAT, employment by sex, occupation and education (Ghana 2025, Nigeria 2025, Kenya 2021, Egypt 2024,\n"
      "South Africa 2025). All ages. Below graduate level = ISCO major groups 4–9 (skill levels 1–2), the ILO's normative measure.")
fig.savefig(FIG / "graduates_below_level.png", dpi=200)
plt.show()

6. Summary numbers for the brief

r25, r15 = rates.xs("25–34", level="age"), rates.xs("15–24", level="age")
summary = {
    **{f"{c}: unemployment 25–34, graduates (%)": half_up(r25.loc[c, "Advanced"]) for c in COUNTRIES},
    **{f"{c}: unemployment 25–34, all (%)": half_up(r25.loc[c, "Total"]) for c in COUNTRIES},
    **{f"{c}: unemployment 15–24, graduates (%)": half_up(r15.loc[c, "Advanced"]) for c in COUNTRIES},
    **{f"{c}: unemployment 15–24, all (%)": half_up(r15.loc[c, "Total"]) for c in COUNTRIES},
    "Egypt: graduate women 25–34 unemployed (%)": half_up(by_sex.loc["Egypt", "Female"]),
    "Egypt: graduate men 25–34 unemployed (%)": half_up(by_sex.loc["Egypt", "Male"]),
    **{f"{c}: graduates' share of employed 25–34 (%)": half_up(grad_share.loc[c, "25–34"]) for c in COUNTRIES},
    **{f"{c}: graduates in office-based jobs (%)": half_up(office.loc[c, "Graduates"]) for c in OCC_COUNTRIES},
    **{f"{c}: all workers in office-based jobs (%)": half_up(office.loc[c, "All workers"]) for c in OCC_COUNTRIES},
    **{f"{c}: graduates in clerical jobs (%)": half_up(grads.loc[c, "4"]) for c in OCC_COUNTRIES},
    **{f"{c}: graduates below graduate level (%)": half_up(below.loc[c, "Below graduate level, total"]) for c in OCC_COUNTRIES},
}
summary = pd.Series(summary)
summary.to_csv(ROOT / "outputs" / "summary_numbers.csv", header=["value"])
summary

Ghana: unemployment 25–34, graduates (%)                10.7
Nigeria: unemployment 25–34, graduates (%)              16.9
Kenya: unemployment 25–34, graduates (%)                11.7
Egypt: unemployment 25–34, graduates (%)                17.7
Senegal: unemployment 25–34, graduates (%)              22.0
South Africa: unemployment 25–34, graduates (%)         21.4
Ghana: unemployment 25–34, all (%)                       4.4
Nigeria: unemployment 25–34, all (%)                     5.9
Kenya: unemployment 25–34, all (%)                       8.1
Egypt: unemployment 25–34, all (%)                       8.5
Senegal: unemployment 25–34, all (%)                     7.8
South Africa: unemployment 25–34, all (%)               37.3
Ghana: unemployment 15–24, graduates (%)                13.2
Nigeria: unemployment 15–24, graduates (%)              18.5
Kenya: unemployment 15–24, graduates (%)                22.8
Egypt: unemployment 15–24, graduates (%)                42.8
Senegal: unemployment 15–24, graduates (%)              29.8
South Africa: unemployment 15–24, graduates (%)         35.8
Ghana: unemployment 15–24, all (%)                       5.9
Nigeria: unemployment 15–24, all (%)                     7.0
Kenya: unemployment 15–24, all (%)                      12.3
Egypt: unemployment 15–24, all (%)                      15.3
Senegal: unemployment 15–24, all (%)                     5.0
South Africa: unemployment 15–24, all (%)               53.2
Egypt: graduate women 25–34 unemployed (%)              31.0
Egypt: graduate men 25–34 unemployed (%)                10.6
Ghana: graduates' share of employed 25–34 (%)           14.1
Nigeria: graduates' share of employed 25–34 (%)         11.7
Kenya: graduates' share of employed 25–34 (%)            7.9
Egypt: graduates' share of employed 25–34 (%)           25.6
Senegal: graduates' share of employed 25–34 (%)          8.7
South Africa: graduates' share of employed 25–34 (%)    18.1
Ghana: graduates in office-based jobs (%)               56.7
Nigeria: graduates in office-based jobs (%)             37.1
Kenya: graduates in office-based jobs (%)               55.4
Egypt: graduates in office-based jobs (%)               69.1
South Africa: graduates in office-based jobs (%)        63.0
Ghana: all workers in office-based jobs (%)              8.2
Nigeria: all workers in office-based jobs (%)            8.0
Kenya: all workers in office-based jobs (%)             10.1
Egypt: all workers in office-based jobs (%)             23.4
South Africa: all workers in office-based jobs (%)      25.5
Ghana: graduates in clerical jobs (%)                    6.8
Nigeria: graduates in clerical jobs (%)                  4.6
Kenya: graduates in clerical jobs (%)                    3.1
Egypt: graduates in clerical jobs (%)                    8.8
South Africa: graduates in clerical jobs (%)            13.4
Ghana: graduates below graduate level (%)               40.1
Nigeria: graduates below graduate level (%)             63.0
Kenya: graduates below graduate level (%)               23.9
Egypt: graduates below graduate level (%)               31.9
South Africa: graduates below graduate level (%)        29.4
dtype: float64

7. Limitations

1. Occupation by education covers all ages. ILOSTAT does not publish it by age, so section 4 describes all employed graduates, not only young ones.
2. Major groups only. Exposure varies within groups; many graduate professionals are teachers and health workers, whose jobs are not exposed. The office-based share is an upper bound.
3. Exposure is not job loss. The tiers measure what generative AI could do to tasks, not what it will do.
4. Unemployment is a narrow measure. It leaves out graduates who have stopped looking for work and those in informal or part-time work below their skills.
5. Survey differences. Years, surveys and classifications differ (Kenya's data are from 2021; South Africa uses ISCO-88), so cross-country comparisons are approximate.
6. Small samples. Few people aged 15–24 have tertiary education, so rates for that group, and especially by sex, are less precise.