Pre-departure playbook — Portugal + Ireland fieldwork

For: Katia Antunes Departure: ~3 weeks (early June 2026) Purpose: A working plan for what to do now, what to do in country, and what kind of specificity to bring into each interview. Plus the EU-vs-national-vs-subnational mental model you need before walking into any of these rooms.

This is a working document. Edit it, mark up the shortlist, scratch out what doesn’t apply. Don’t treat it as a textbook chapter.

The short version

Three weeks is enough to do this well if you start today. The mistake junior researchers make on a first fieldwork trip is showing up with a vague “I’m interested in rural blended finance, can you tell me about your work?” — which produces vague answers, polite meetings, and zero data. The fix is to walk in with a sharply-defined research question, a specific data ask, and an offer of something in return.

Three weeks gives you: - Week 1 (this week): sharpen the question, build the stakeholder list, write the first wave of cold/warm emails. - Week 2: schedule interviews, read the public reports of the organizations you’re meeting, draft an interview guide. - Week 3: finalize logistics, prepare per-meeting briefing notes, send confirmation emails 48h before each meeting.

In country you do 12-25 interviews across Portugal and Ireland combined. You take notes in real time. Within 24 hours of every interview you write a 1-page memo: who, what they said, what data exists, what they offered, what to follow up on. That memo file becomes your dissertation appendix.

What to do this week

1. Sharpen the research question down to one sentence

Right now your topic is “rural blended finance in Portugal” or thereabouts. That is a topic. It is not a question. Walk into BPF or AD&C with a topic and you will get a 30-minute tour of their website. Walk in with a sentence-long question and you will get data.

Try the sentence-completion exercise. Fill in:

“I want to know whether [specific intervention / instrument] in [setting] affects [outcome] for [unit], identified by [design].”

Three candidate sharpenings, ranked roughly by feasibility:

The take-up question. “I want to know whether the SBCI / BPF-backed credit lines reach rural SMEs at the same rate as urban SMEs, identified by an RDD on the NUTS-3 GDP-per-capita eligibility cutoff for regional aid intensity.” This is the cleanest research design, leverages the EU regional-aid 75%-of-EU-average threshold (a known sharp cutoff), and BPF / SBCI both lend through commercial-bank intermediaries which makes them comparable.
The targeting question. “I want to know whether LEADER / GAL allocations within Portugal flow to the most-disadvantaged rural areas, identified by a DiD using the 2014→2020→2023 LEADER reallocation rules.” This is harder because LEADER program rules change over funding cycles, but the data exist (IFAP publishes recipient-level disbursements) and the variation is legitimate.
The distributional question. “I want to know whether rural-credit beneficiaries see income gains concentrated in the bottom of the rural household distribution, identified by quantile regression on INE household-survey data linked to credit registry data.” This is the most ambitious because it requires merging two data sources, but it makes the policy-relevant inequality argument directly. The quantile-regression chapter of your bootcamp is the spine.

Pick ONE. Bring it into every interview. If an interviewee can’t speak to it, ask “who would?” That’s how the snowball works.

2. The stakeholder map — read alongside `portugal_stakeholders.md` and `ireland_stakeholders.md`

Detailed in the sibling files. The condensed version:

Portugal first-call shortlist (email this week): 1. AD&C (national managing authority for EU funds). They architect the whole PT 2030 structure. If they will share unit-level disbursement data, your dissertation has a spine. 2. BPF (national promotional bank). They issue most national-level blended-finance products. They can tell you the application rules, the rejection rates, the rural carve-outs. 3. GPP (the agriculture-ministry policy/evaluation unit). They run impact evaluations on CAP and may have a co-author appetite. 4. Banco de Portugal Economics Research. They have a credit register and have published rural-credit working papers. If they will let you query an aggregated extract, that’s a paper by itself. 5. Católica-Lisbon SBE (relevant faculty in finance + dev econ). One academic ally inside Portugal opens 5 more doors. Identify which faculty have worked on rural credit and email cold with your one-sentence research question.

Ireland first-call shortlist: 1. ESRI. They are the Irish evaluation body. An ESRI affiliation as a visitor (a desk + a coffee budget) for two weeks is a career-grade credential. 2. SBCI. The Irish parallel to BPF. They will know which commercial banks pass SBCI credit through and which rural areas show low takeup. 3. DRCD (Rural Development department). They run RRDF and LEADER. They have geographic disbursement data. 4. ISIF. Sovereign development finance, regional-investment mandate. Different model than the EU-pass-through Portuguese pattern, which makes Ireland a useful comparison case. 5. UCD Geary Institute or Teagasc Rural Economics. Same logic as Católica-Lisbon — one academic ally opens five more doors.

3. Write the first-wave emails — TODAY

A first-wave cold/warm email should be 4 sentences:

Dear [Name],

I’m a researcher working with Dr. Ian Helfrich (Georgia Tech PhD, 2024, NMTC blended-finance work) on a project about whether rural-targeted blended finance reaches the populations the programs are designed to serve. I’ll be in Lisbon from [date] to [date], with research interest in the design and evaluation of the PT 2030 / EAFRD instruments your team operates. Could I request 30 minutes of your time to learn from your perspective on what the evidence base looks like and what questions you think the research community is missing? I’d be glad to share my findings draft when it’s ready as a working-paper contribution.

Best, Katia Antunes

Notes: - The “with Dr. Helfrich” sentence is leverage. He is a known quantity on NMTC. Use it. - “30 minutes” is the magic number. Easier yes than 60. They’ll usually take 50 anyway. - The “share my findings” sentence promises reciprocity. Without it, you’re asking for free time. - Use the role title from the stakeholder map if you don’t know an individual name. “Dear Head of Programme Evaluation,” is fine for a first round.

Send 12-15 of these on Monday. Expect 5-8 responses. Set up the 5-8 that respond as your week-2 calendar anchors.

4. Start the reading

For each organization you’ll meet, you must read their annual report or their two most recent published evaluations before the meeting. Not just the home page. The actual PDF. This takes 30-45 minutes per org. Do it before the meeting, not on the plane.

Put a 1-page brief on each org in a single shared Drive folder or a Notion page: - What they do, in your words (one paragraph) - One question you have about something specific in their published work - The data you want to ask about - The person you’re meeting

This brief is what you read on the metro the morning of the meeting. It is the difference between “interesting chat” and “they offered to share disbursement data”.

What kind of specificity does each interview need

The right level of specificity depends on who you’re talking to. Five categories, rough decreasing order of specificity:

Category A: Operational / loan-officer / programme-implementation staff

Example: A credit officer at Crédito Agrícola, a programme manager at a GAL, a loan-disbursement analyst at IFAP, an SBCI underwriting manager.

Specificity level: very high. You should know: - The product name(s) they administer - The eligibility rules in detail (criteria, thresholds, application paperwork) - The default rates published in their most recent report - The geographic distribution of disbursement if public

Ask: “Walk me through the application process for [specific product]. Where do most applications fall off?” Get the attrition funnel numbers if you can. Ask whether decisions are rule-based or discretionary at the cutoff. The cutoff geography is your RDD; the discretion at the cutoff is your identifying-assumption story.

Category B: Policy designers at the ministry / managing-authority level

Example: AD&C senior staff, DRCD policy unit, GPP evaluators, DAFM CAP planning.

Specificity level: high but more conceptual. They will speak in programme acronyms. Master those before the meeting — at minimum: PT 2030, PEPAC, EAFRD, ERDF, ESF, CSP (CAP Strategic Plan), JEREMIE/JESSICA legacy instruments, InvestEU. Don’t make them define acronyms for you.

Ask: “What’s the theory of change for [specific instrument]? What evidence base did the 2023-2027 design draw on? What evaluation is planned?” Get them to articulate the gap between what evidence exists and what the program assumes. That gap is your research contribution.

Category C: DFI / promotional-bank product owners

Example: BPF product managers (rural / SME / cohesion-fund products), SBCI product team, EIB country desk.

Specificity level: high on the product, medium on the broader programme. They know their book. They know rejection rates and average loan sizes and default patterns. They are also institutionally cautious about sharing — your “what can I get” question has to be specific.

Ask: “If I were to write a paper on take-up of [specific product] in rural vs urban Portugal, what would the ideal data look like, and what part of that is publicly available?” This forces them to triage: what’s already public (point you there), what they could maybe share under a research agreement (now you have a path), what is hard-confidential (acknowledge and move on).

Category D: EU-presence offices

Example: EIB Lisbon, EIB Dublin, EIF, European Commission Representations.

Specificity level: medium. They know the EU-side rules and the cross-country comparisons. They are less specific about the domestic implementation.

Ask: “How does the Portuguese / Irish implementation of [EAFRD or InvestEU or RRF] compare to peer member states in terms of [specific dimension]? What’s working, what’s stuck?” These offices love being asked comparative questions because the EU-side comparative perspective is their actual expertise.

Category E: Academia and think tanks

Example: ESRI, Católica-Lisbon, UCD Geary, ISEG, Teagasc.

Specificity level: medium-high on methodology, high on the literature. They expect you to know the methodological landscape and to bring a specific research question. They are the most likely to want a back-and-forth conversation rather than to be “interviewed”.

Ask: “I’m framing the take-up question as an RDD on the regional-aid threshold. The McCrary density test will probably reject because regions self-select into the 75%-of-EU-average bucket through GDP-revision strategy. What’s your read on whether that design is salvageable?” This signals you read the methods literature; they will engage at peer level rather than mentor-teaching level.

In Category E meetings, bring a draft. Even 3 pages of “here’s my research question, here’s the identification strategy, here’s the data I need” is enough. Academic interviewees give the best feedback when they have something concrete to push back on.

The 60-second pitch you have to memorize

Before any meeting you may be asked “so what are you working on?” You need a 60-second answer that lands. Write it, edit it, say it out loud to yourself three times before you leave the apartment.

The structure:

Who I am (8 seconds). “I’m Katia Antunes, a researcher working with Dr. Ian Helfrich at Georgia Tech.”
What the question is (12 seconds). “I’m studying whether rural-targeted blended-finance instruments actually reach the populations they’re designed to serve, with a focus on Portugal and Ireland as EU comparison cases.”
What’s new about how I’m doing it (15 seconds). “I’m extending a US-context methodology (Helfrich 2026 on the New Markets Tax Credit) to the EU rural-finance setting, using the regional-aid eligibility cutoffs and LEADER programme variation as identification.”
The ‘why I’m here’ (15 seconds). “Right now I’m trying to map the institutional landscape and identify data sources. I’d love to learn from your team about [specific thing they do].”
The reciprocity offer (10 seconds). “I’m happy to share working-paper drafts, code, or a results brief once the analysis is further along.”

Total: 60 seconds. Edit until it lands cleanly without filler.

The EU-local vs national-level vs subnational distinction

This is the conceptual frame you need before any of these interviews make sense. Three layers, four flows, plenty of friction.

The three layers

Layer 1 — EU-level. Brussels designs the rules and the multi-year financial framework (MFF). The big rural-relevant instruments are: - EAFRD (European Agricultural Fund for Rural Development) — CAP Pillar 2. The rural-development pillar. €87B EU-wide over the 2023-2027 CAP. - ERDF (European Regional Development Fund) — funds regional cohesion projects including some rural-SME finance. - ESF (European Social Fund) — funds social inclusion, sometimes including financial literacy / cooperative-banking projects. - InvestEU — the EU’s financial-instruments umbrella that succeeds JEREMIE/JESSICA/etc. Mostly delivered through EIF guarantees and EIB lending to national intermediaries. - RRF (Recovery and Resilience Facility) — the post-COVID instrument, includes rural-development envelopes.

Brussels does not directly lend to a Portuguese farmer or an Irish SME. They allocate envelopes to member states.

Layer 2 — National-level (member state). The national managing authority converts the EU envelope into national programmes: - Portugal: AD&C is the managing authority for ERDF/ESF flows under PT 2030. IFAP is the paying agency for CAP/EAFRD. PEPAC (Plano Estratégico da PAC) is the national CAP Strategic Plan. BPF / IFD legacy products are the national-promotional-bank delivery vehicles for some of this. - Ireland: DRCD operates LEADER and RRDF. DAFM is the paying agency for CAP. SBCI is the national promotional bank for SME finance. The Irish CSP is the equivalent of PEPAC.

National layer adds national rules — eligibility thresholds, application processes, intermediary selection. Same Euro, different design constraints in each member state.

Layer 3 — Subnational / regional / local. The money flows from national managing authority to: - Regional managing authority (in Portugal: the 5 CCDRs; in Ireland: weaker, more centralised — county councils and LAGs play this role) - Local action groups (LEADER GALs in Portugal, LAGs in Ireland — these are public-private partnerships that allocate small grants and credits within their territory) - Local intermediary (a rural cooperative bank, a local enterprise office, a municipal credit fund) - Final beneficiary (a farmer, a rural SME, a community project)

The four flows that matter for identification

When you sit in a BPF or SBCI meeting, you are looking at the seam between Layer 2 (national) and Layers 1 (EU) and 3 (local). Four flow types create econometric opportunities:

EU-to-national allocation rules. Member states get more EAFRD per capita if they have lower GDP / more rural population / weaker prior outcomes. The allocation formula is mechanical and known. This gives you cross-country variation that is plausibly exogenous to country-specific shocks within a single allocation cycle.
National-to-regional allocation rules. Portugal’s CCDR-level allocations are formula-driven within PT 2030 (population, rural population, GDP gap from EU average). Ireland is more centralised but LEADER allocations per LAG depend on population and rurality indicators. The formulas are RDD candidates if there are thresholds (Portugal’s NUTS-3 eligibility cutoff at 75% of EU GDP) and DiD candidates if there are reform years that shift the formula.
Local-eligibility cutoffs. A rural cooperative bank lending under a BPF-guaranteed product applies a “rural area” definition (e.g., low-density municipality, parish-level deprivation index, distance from district capital). Those definitions are sharp cutoffs at the local level. Sharp cutoffs are RDDs.
Cycle-to-cycle reform. EU CAP cycles run 7 years. 2007-2013, 2014-2020, 2023-2027. Each transition redesigns rules. The transition years are quasi-experimental treatment dates if the reform is plausibly orthogonal to local rural-finance demand (and it often is, because the redesign is driven by Brussels-level political horse-trading).

What this means for your interviews

At Layer 1 institutions (EIB, EIF, EC Reps) you ask about the allocation rules and their political economy. They know the cross-MS comparison.
At Layer 2 institutions (AD&C, BPF, SBCI, DRCD, DAFM) you ask about the design choices and the within-country geographic disbursement data. They know the national rules.
At Layer 3 institutions (CCDRs, GALs, LAGs, local cooperative banks) you ask about the application process and the actual filtering of who gets funded. They know the operational reality.

A good fieldwork trip touches all three layers. A bad one is all Layer 2 interviews, which gives you a generic policy overview and no identification.

EU-local vs national-level — when does the distinction matter for the paper?

For your eventual paper, the EU-vs-national distinction matters in three places:

(a) The treatment definition. “Treated” can mean: - The municipality crossed the EU regional-aid threshold (EU-defined, mechanical) - The municipality is in a BPF-eligible rural carve-out (national-defined, possibly discretionary) - The municipality is in a LAG with LEADER allocation above some level (local-defined, soft)

The first is the cleanest RDD. The second is fuzzier. The third is closer to DiD on rollout timing.

(b) The funder audience. If you eventually pitch this to a Gates / IFC / EIB officer, they want to know “is this EU money working” or “is this national money working” — those are different policy levers. Be explicit about which one your treatment captures.

(c) The external validity. A finding on EU-rule variation generalises across MS (other EU countries have the same rule). A finding on Portuguese-specific national-program variation generalises only to Portugal. Be honest about which you’re claiming.

A 1-page checklist for each meeting

Org name: ________________________
Date / time: ______________________
Where: ___________________________
Who I'm meeting: __________________
Their role: _______________________

Before:
[ ] Read their two most recent annual reports / evaluations
[ ] Identify 2-3 specific publications by people I might be meeting
[ ] Have one specific question about something in those publications
[ ] Confirm 48 hours before (and 24 hours before)
[ ] Print or save my one-page brief on this org
[ ] Pack: laptop, notebook, two pens, business cards if you have them

During:
[ ] 60-second pitch, edited
[ ] My 4-5 core questions
[ ] Note-taking: paraphrase + quoted phrases where the wording matters
[ ] Ask: what data is publicly available; what isn't; what's the path to research access
[ ] Ask: who else should I be talking to (snowball)
[ ] Ask: can I follow up by email with specific data requests

After (within 24 hours):
[ ] 1-page memo: who, what they said, what data, what follow-up
[ ] Thank-you email (same day if possible)
[ ] Any specific follow-up email with the data request, citing what they offered
[ ] Update the running 'orgs and data' tracker
[ ] If snowball name was offered: send a cold email naming the referrer ("X suggested I reach out to you")

The thank-you email template

Same-day, short:

Dear [Name],

Thank you for the conversation today. The points you made about [specific thing they said] gave me a much clearer picture of how [specific topic] operates in practice, especially [specific detail]. I’ll follow up next week with a more specific request about [the data / contact / document they offered]. If it’s helpful, I’ll share my draft findings when they’re further along.

Best, Katia

The specific-thing-they-said line is the part that matters. It signals you were listening, you noted what was new to you, and you’re not generic-spamming. People remember being listened to.

What to bring back

By the end of the trip you should have:

A folder of 1-page memos, one per interview. ~15-25 of them.
A running data-tracker spreadsheet: org, contact, data offered, status (asked, pending, received, no), what format, what level of aggregation.
A stakeholder snowball map: who knew who, who recommended whom. This becomes a network diagram in your dissertation.
A revised research-question doc: the one-sentence question you started with, plus how it has evolved based on what people told you about the data and the institutions.
A shortlist of 3 papers worth writing out of the trip — your primary dissertation paper, a methodological note, and one short comparative descriptive piece.
At least two ongoing research relationships — one with a Portuguese institution, one with an Irish institution — that will let you ask follow-up questions after you’re home.

If you come back with all six, the trip was a success.

What you should NOT try to do on this trip

Don’t try to write the paper in-country. You can’t think clearly while jet-lagged and shifting between languages.
Don’t try to run regressions on data you receive in-country. Bring it home. Run it carefully. The whole point of the trip is access, not execution.
Don’t make any data agreements without forwarding them to Dr. Helfrich first. Some institutions will offer access under terms that look fine but have IP clauses that complicate publication.
Don’t burn the snowball. Every referral is an investment. Send the thank-you, follow up on the data ask, and close the loop with the referrer (a one-line “I met X, very helpful, thanks for the intro” goes a long way).
Don’t promise authorship to anyone you meet without checking first. Authorship norms in EU institutional research differ from US academic norms, and a casual “we should write something together” said in a meeting can become a binding obligation in the institutional context.