Chapter Measuring the movement between employment and self-employment: a survey proposal

This is not an experimental study but a methodological proposal and descriptive analysis. The author tested:

**Intervention:** A proposed set of 12–15 new survey questions designed to capture the nuances of self-employment transitions (e.g., reasons for switching, type of self-employment, duration of spells, income volatility).

**Comparator:** The existing Italian National Institute of Statistics (Istat) labour force survey questions, which the author argues are too coarse to distinguish between genuine entrepreneurship, freelance gig work, and disguised employment.

**Outcome measures:** The ability of the proposed questions to classify workers into finer categories (traditional craftspeople, freelancers, learned professionals, dependent self-employed) and to track year-to-year movements between employment and self-employment.

The paper also analyses secondary data from Istat's continuous labour force survey (RCFL) and the European Union Labour Force Survey (EU-LFS) to describe trends in self-employment composition from 2009 to 2019.

**Sample size:** Approximately 5.5 million self-employed workers in Italy (official estimate for the period), drawn from a national labour force survey sample of roughly 175,000 households per quarter (the standard Istat RCFL sample).

**Population:** Italian residents aged 15–74 who reported being employed or self-employed in the reference week of the survey.

**Setting:** Nationwide, covering all regions of Italy, with data collected via face-to-face and telephone interviews. The analysis spans 2009–2019, with a brief extension to 2020 (Covid-19 year).

**Subgroups examined:** Younger workers (aged 15–34), older workers (aged 50+), educational attainment levels (less than high school, high school, university degree), and occupational categories (craftspeople, shopkeepers, freelancers, professionals).

The paper uses existing survey instruments and proposes new ones:

**Existing instrument:** Istat's Labour Force Survey (RCFL) — a quarterly household survey harmonised with Eurostat definitions. Key variables include: employment status (employee vs. self-employed), occupation (ISCO codes), economic sector (NACE codes), hours worked, income bracket (categorical), and reason for leaving previous job.

**Proposed instrument:** A new module of 12–15 questions covering:

- Type of self-employment (sole proprietor, partner in a firm, freelance professional, gig worker, dependent self-employed)

- Number of clients in the past year (single client vs. multiple clients — a key indicator of disguised employment)

- Decision-making autonomy (e.g., "Can you decide your own working hours?" on a 1–5 scale)

- Income stability (e.g., "How many months in the past year did your income fall below €1,000?" — a continuous measure)

- Reason for entering self-employment (voluntary vs. involuntary: "Did you choose self-employment because you could not find a paid job?" — binary)

- Duration of self-employment spell (months)

- Previous employment status (employee, unemployed, inactive)

**Data analysis:** Descriptive statistics (percentages, absolute counts) and trend analysis using linear interpolation for missing years. No inferential statistics (no p-values, confidence intervals, or effect sizes) are reported — the paper is purely descriptive and methodological.

**Study design:** This is a methodological paper combined with a descriptive trend analysis. It is not an experiment, randomised trial, or quasi-experimental study. The author reviews existing survey questions, identifies gaps, proposes new questions, and then uses existing survey data to illustrate why the new questions are needed.

**Data source:** The Istat RCFL is a continuous survey with a rotating panel design. Each household is interviewed for two consecutive quarters, then skipped for two quarters, then interviewed again for two quarters (a 2-2-2 pattern). This allows for some longitudinal tracking of individuals across 15 months, but the paper primarily uses cross-sectional data (yearly averages).

**Duration:** The analysis covers 11 years (2009–2019) of yearly data, plus a brief look at 2020. The proposed survey questions are not tested in the field — they are a theoretical proposal.

**Randomisation:** Not applicable. The RCFL uses a stratified multi-stage probability sample of households, but the paper does not describe the sampling design in detail.

**Blinding:** Not applicable. Survey respondents are not blinded to questions, and interviewers are not blinded to the study purpose.

**What this design can and cannot prove:**

**Can prove:** The paper can describe trends in self-employment composition over time (e.g., the decline in young, less-educated entrants). It can identify gaps in current measurement tools. It can propose a more granular classification system.

**Cannot prove:** Causality — the paper cannot say *why* the trends occurred (e.g., whether the decline in young entrants was due to labour market policies, economic recession, or cultural shifts). It cannot validate the proposed survey questions (no pilot testing, no reliability or validity statistics). It cannot estimate the true size of the "dependent self-employment" phenomenon (workers who are legally self-employed but functionally employees) because the current survey does not capture it.

**Major methodological weaknesses:**

1. **No validation of proposed questions:** The author suggests new questions but provides no data on how they perform (e.g., test-retest reliability, convergent validity against administrative records, response rates).

2. **Descriptive only:** No statistical tests for trends (e.g., no regression analysis to determine if the decline in young entrants is statistically significant).

3. **Single country:** Italy has a unique labour market with a large informal sector and strong professional guilds. Findings may not generalise to other countries.

4. **Covid-19 year treated superficially:** 2020 data is mentioned but not systematically analysed.

5. **No disaggregation by gender or region:** The paper notes these are important but does not present the data.

The paper reports descriptive trends from the Istat data. All numbers are absolute counts or percentages from the survey sample, extrapolated to the national population.

**Overall self-employment size:** Approximately 5.5 million workers in Italy (about 22% of total employment) during the 2009–2019 period, relatively stable in absolute terms.

**Decline in young entrants (aged 15–34):** The number of self-employed workers aged 15–34 fell from approximately 1.4 million in 2009 to 0.9 million in 2019 — a decline of about 500,000 workers (a 36% drop). This was driven primarily by less-educated workers (high school diploma or less).

**Increase in older entrants (aged 50+):** The number of self-employed workers aged 50+ rose from approximately 1.2 million in 2009 to 1.6 million in 2019 — an increase of about 400,000 workers (a 33% rise). This was driven primarily by highly educated workers (university degree).

**Educational shift:** Among self-employed workers aged 15–34, the proportion with a university degree rose from 12% in 2009 to 22% in 2019. Among those aged 50+, the proportion with a university degree rose from 8% to 15%.

**"Disappeared" one million:** The author calculates that the net loss of younger, less-educated self-employed entrants (about 1 million people) was only partly compensated by the gain of older, highly educated entrants (about 400,000 people), leaving a net deficit of roughly 600,000 workers who exited self-employment without being replaced.

**Occupational composition:** Traditional craftspeople and shopkeepers declined from about 60% of self-employment in 2009 to 50% in 2019. Freelancers and professionals (e.g., lawyers, consultants, IT contractors) rose from 25% to 35%. Gig workers and dependent self-employed (e.g., delivery drivers, platform workers) emerged as a new category, estimated at 5–10% of self-employment by 2019, though the current survey cannot accurately count them.

**Involuntary self-employment:** Among new entrants in 2019, approximately 30% reported entering self-employment because they could not find a paid job (based on a single survey question). This proportion was higher among younger workers (40%) than older workers (20%).

**Income volatility:** Among self-employed workers, about 25% reported income below €1,000 per month for at least 3 months in the past year. This was more common among younger workers (35%) and gig workers (50%).

**Primary vs. secondary outcomes:** The paper has no formal primary outcome. The main descriptive finding is the compositional shift from young, less-educated to older, highly educated self-employed. The secondary finding is the inadequacy of current survey questions to capture this shift.

Because this is a descriptive study with no experimental manipulation, there are no effect sizes in the traditional sense. However, the magnitude of the demographic shift is large in practical terms:

**A 36% decline in young self-employed entrants over 10 years** is equivalent to roughly 50,000 fewer young people entering self-employment each year. For context, this is about the size of the entire graduating class of a mid-sized Italian university.

**A 33% increase in older self-employed entrants** means that by 2019, nearly 1 in 3 self-employed workers was over 50, compared to 1 in 4 in 2009.

**The educational shift** means that a young person with a university degree is now about twice as likely to be self-employed as one without (22% vs. 12% of the age group), whereas in 2009 the gap was much smaller.

**The "missing million"** — the net loss of young, less-educated workers from self-employment — represents about 18% of the total self-employed population. If this trend continues, the self-employed workforce will age significantly, with potential consequences for innovation, entrepreneurship, and social security contributions.

**Acknowledged by the author:**

The proposed survey questions have not been pilot-tested or validated.

The Istat data may undercount certain types of self-employment (e.g., gig workers who do not identify as self-employed).

The analysis is limited to Italy; findings may not apply to other countries with different labour market structures.

The Covid-19 year (2020) is treated as an outlier and not systematically integrated into the trend analysis.

The paper does not control for macroeconomic factors (e.g., GDP growth, unemployment rate, tax policy) that could explain the trends.

**Critical reader notes:**

**No statistical significance testing:** The paper reports raw numbers and percentages but never tests whether the observed trends are statistically significant. A decline of 500,000 workers over 10 years could be within sampling error, though the magnitude makes this unlikely.

**No confidence intervals:** The reader cannot assess the precision of the estimates. The Istat sample of 175,000 households is large, but subgroup estimates (e.g., young, less-educated self-employed) may have wider margins of error.

**Single data source:** The paper relies entirely on self-reported survey data. Administrative data (e.g., tax records, social security registrations) could provide a more accurate count but is not used.

**No gender or regional analysis:** The paper notes that self-employment rates vary by gender (higher for men) and region (higher in the South), but does not break down the trends by these factors. This is a significant omission because the decline in young entrants may be concentrated in specific regions or genders.

**Definitional ambiguity:** The paper proposes new definitions but does not resolve the fundamental problem of how to classify workers who hold multiple jobs (e.g., part-time employee and part-time freelancer). The current survey forces a single primary status.

**No longitudinal tracking:** The paper uses cross-sectional data (different people each year), so it cannot track individual transitions. The proposed survey questions would improve this, but the paper does not test them.

**Publication bias:** The paper is a chapter in a book (likely an academic monograph), not a peer-reviewed journal article. The review process may have been less rigorous.

For someone running their own n=1 experiment on career transitions (e.g., moving from employment to self-employment), this paper offers a framework for what to measure and what confounds to watch for. However, because the paper is a methodological proposal and not an experimental study, the takeaways are about measurement design rather than specific interventions.

### What to test

**Your own transition type:** Are you moving to self-employment voluntarily (e.g., pursuing a passion) or involuntarily (e.g., laid off and unable to find a job)? The paper suggests this distinction matters for outcomes like income stability and satisfaction.

**Your self-employment category:** Are you a freelancer (multiple clients, high autonomy), a dependent self-employed worker (single client, low autonomy), or a traditional business owner? The paper shows these categories have very different income trajectories.

**Your income volatility:** Track how many months per year your income falls below a threshold (e.g., €1,000/month or your minimum living expenses). The paper found that 25% of self-employed workers experience this.

### Minimum meaningful duration

**At least 12 months:** The paper's trend analysis covers 11 years, but for an individual experiment, 12 months is the minimum to capture seasonal income fluctuations and the full cycle of client acquisition and retention.

**Quarterly check-ins:** The paper uses quarterly survey data. For your own experiment, measure income, hours worked, and satisfaction every 3 months to spot trends.

### What to measure (specific metrics)

Based on the paper's proposed survey questions, track these metrics in a spreadsheet or journal:

1. **Income stability:**

- Monthly gross income (€)

- Number of months with income below €1,000 (or your personal threshold)

- Number of clients per month (single client = higher risk of disguised employment)

2. **Autonomy:**

- "Can you decide your own working hours?" (1 = never, 5 = always)

- "Can you choose which projects to accept?" (1–5 scale)

3. **Reason for transition:**

- "Did you choose self-employment because you could not find a paid job?" (yes/no)

- "Did you choose self-employment to pursue a specific opportunity?" (yes/no)

4. **Duration:**

- Months since you started self-employment

- Expected duration (do you plan to continue for >2 years?)

5. **Well-being:**

- Work satisfaction (1–10 scale)

- Stress level (1–10 scale)

- Hours worked per week (compare to your previous employment)

### Key confounds to control for

**Macroeconomic conditions:** The paper shows that self-employment trends are influenced by the overall economy (e.g., the 2008 recession, Covid-19). If you run your experiment during a recession, your results may not generalise to boom times.

**Age and education:** The paper found that younger, less-educated workers fare worse in self-employment. If you are under 35 or lack a university degree, expect higher income volatility.

**Industry:** Traditional crafts and retail declined, while professional services grew. If you are in a declining industry, your transition may be harder.

**Single-client dependency:** The paper highlights that workers with a single client (e.g., Uber drivers, some consultants) are functionally employees but lack protections. If you have one main client, track how much of your income comes from them.

**Tax and social security:** Italy has specific tax regimes for self-employed workers (e.g., the "forfettario" flat tax). Your country's tax system will affect your net income.