---
title: "What a Macro Analyst Needs: A Sourced Competency Map"
description: "Communication, data handling and disciplined judgement recur in the UK economists' standard, Bank of England and IMF reviews, and an employer survey."
url: "https://staging.khoi.co.uk/en/blog/nha-phan-tich-vi-mo-can-gi"
lang: en-GB
author: "Lê Quốc Khôi"
date: 2026-11-03
updated: 2026-11-03
topics: ["Data & analytics"]
tags: ["Macro analyst", "Economic forecasting", "Professional competencies", "Self-study plan"]
---

# What a macro analyst needs: a competency map from public sources

## Key takeaways

- The UK government economists' standard splits competence into three areas: application of knowledge, analysis of data and effective communication (Government Economic Service, 2022, p. 4). Communication to non-specialists recurs in more sources than anything else; 75.6% of respondents in a survey of 40 employers rated it 'very' important (Lait and Lane, 2019, p. 3).
- Judgement is one of the few areas with measurements. At the IMF, country-specific experience was associated with smaller GDP forecast errors for low-income countries but showed little relation for advanced and emerging economies (2007 to 2011) (Independent Evaluation Office of the International Monetary Fund, 2014, p. 23).
- In an IARPA-sponsored forecasting tournament, 60 people selected on accuracy stayed highly accurate for two years running. The authors report frequency of belief updating as the strongest single behavioural predictor of accuracy, citing an earlier paper by the same group (Mellers et al., 2015, p. 267, 277). The questions were mainly geopolitical, not macro series.
- On tools, IMF country desks used spreadsheet frameworks and judgement more than structural econometric models (Independent Evaluation Office of the International Monetary Fund, 2014, p. 13); large models such as DSGE sit in the central-bank toolkit, kept up by dedicated teams (Bernanke, 2024, p. 7, 16).
- No source I read tested whether people with the full list analyse better, and none is about Vietnam. The post marks which items have evidence and which are my synthesis.

Who gets to say what a macro analyst should know? Usually someone with a job advert or a course to sell. At the Bank of England, a little over half of the people working on the forecast hold master's-level qualifications and about a third hold doctorates (Bernanke, 2024, p. 20), which a self-learner might read as a bar to clear. I would rather ask what the institutions that employ forecasters, and the researchers who study them, have put in writing, and keep my own additions visibly apart. The study plan near the end is one of those additions.

## What evidence exists on a macro analyst's skills, apart from job adverts?

Six public documents, each viewing the job from a different side. Job adverts are out: they show what an employer wants, not who does the work well. I also searched for a Vietnamese counterpart from the State Bank or the General Statistics Office. What came back was securities-firm adverts and a grade regulation for statisticians, not a competency framework for macro analysts, so there is no Vietnamese source here.

| Source | What it is |
|:--|:--|
| Government Economic Service, GES (2022) | Skills standard for UK government economists, built from focus groups and interviews with members in 2021 (Government Economic Service, 2022, p. 3) |
| Bernanke (2024) | Independent review of forecasting at the Bank of England: some 60 interviews and a comparison with six other central banks (Bernanke, 2024, p. 3, 4) |
| IMF Independent Evaluation Office (2014) | Survey and interviews of country desk economists, plus a regression of GDP forecast errors on experience (Independent Evaluation Office of the International Monetary Fund, 2014, p. 23) |
| Mellers and colleagues (2015) | IARPA-sponsored forecasting tournament, 2011 to 2014, with 2,200 to 3,900 forecasters a year (Mellers et al., 2015, p. 268) |
| Lait and Lane (2019) | Survey of 40 employers of economics graduates, mostly central government, consultancy and public services (Lait and Lane, 2019, p. 2) |
| QAA (2023) | UK benchmark statement for economics degrees (Quality Assurance Agency for Higher Education, 2023, p. 3) |

The limits of each are under 'Going deeper'.

## Which items on the map have the most sources behind them?

Communication, data handling and disciplined judgement. The last column counts independent documents that make the point: 'many' is three or more, 'some' is two, 'few' is one. A count shows how many sources mention an idea, not how right the idea is.

| Competency | What the sources say | Level |
|:--|:--|:--|
| Communicating with non-economists | GES makes it one of three areas and calls it 'critical' (Government Economic Service, 2022, p. 3, 4); 75.6% of employers rate it very important (Lait and Lane, 2019, p. 3); QAA (Quality Assurance Agency for Higher Education, 2023, p. 4); Bernanke devotes four recommendations to communication (Bernanke, 2024, p. 10, 11, 12) | Many |
| Understanding data: how built, how late, how revised, how available | GES (Government Economic Service, 2022, p. 10); macro data arrive late, noisy and revised (Bernanke, 2024, p. 13); data availability is the single most important factor in choosing a method (Independent Evaluation Office of the International Monetary Fund, 2014, p. 13) | Many |
| Quantitative and statistical skills | 73.2% of employers rate them very important (Lait and Lane, 2019, p. 3); GES (Government Economic Service, 2022, p. 10, 12); QAA (Quality Assurance Agency for Higher Education, 2023, p. 12) | Many |
| Disciplined judgement, updating beliefs | (Bernanke, 2024, p. 17); (Independent Evaluation Office of the International Monetary Fund, 2014, p. 16, 23); (Mellers et al., 2015, p. 277) | Many, some measured |
| Forecasting and learning from errors | (Bernanke, 2024, p. 13); (Independent Evaluation Office of the International Monetary Fund, 2014, p. 25); simple benchmarks (Hyndman and Athanasopoulos, 2021) | Some |
| Scenarios and conditional forecasts | (Bernanke, 2024, p. 10, 14); sensitivity analysis (Government Economic Service, 2022, p. 12) | Some |
| Core economics and knowing the limits of theory | (Government Economic Service, 2022, p. 4, 6); (Quality Assurance Agency for Higher Education, 2023, p. 15, 16) | Some, expectations only |
| Systems, anticipatory and strategic thinking | (Quality Assurance Agency for Higher Education, 2023, p. 6), following sustainable-development guidance | Few |

The first four rows have at least three documents pointing the same way. The last row holds the kind of label that turns up in many competency lists, and it has the thinnest evidence.

## Mindset: what does 'disciplined judgement' mean once it is measured?

It means scoring your judgement the way you would score a hypothesis, instead of trusting a hunch. Two sources with measurements lean that way, and each has a clear limit.

Start with the IMF, where the gap is easiest to see. Only half of the desk economists who answered the evaluation's survey said that an analysis of past forecast performance had been done on their desk during their tenure (Independent Evaluation Office of the International Monetary Fund, 2014, p. 25). Box 4 regresses GDP forecast errors on experience: more experience goes with lower errors, but country-specific experience shows clearly only for low-income countries (Independent Evaluation Office of the International Monetary Fund, 2014, p. 23). The report offers a possible explanation: those countries have little data and few outside forecasters, so judgement is used more. It is a correlation in internal data, not an experiment.

The tournament evidence adds behaviour. At the end of each year Mellers and colleagues (2015) picked 60 forecasters on accuracy and put them in elite teams. They did not drift back to average; their area under the ROC curve, a measure of separating signal from noise, was 96%, against 84% and 75% in the comparison groups (Mellers et al., 2015, p. 269, 270). In year two they updated 5.64 times per question, against 2.15 and 1.79 (Mellers et al., 2015, p. 274). The measured traits and habits correlated with accuracy only weakly, between 0.07 and 0.23 in size (Mellers et al., 2015, p. 275).

Bernanke supplies the institutional view. No central bank relies on models alone; human judgement is a critical, sometimes dominant, part of the forecast (Bernanke, 2024, p. 17). He also warns that making each forecast a small revision of the last, topped up with ad hoc judgements, can hide deeper problems, and he wants staff charged with flagging large errors and their sources (Bernanke, 2024, p. 9). Inflation from mid-2021 was largely, though not entirely, missed by every central bank in his comparison (Bernanke, 2024, p. 38). Professional teams missed it, so the habit to copy is how they handle error.

My inference from the three: what self-study can change is habit, not native ability. Writing a forecast and its reasons down, scoring it later and updating when news arrives can all start this week.

## Skills: why does communication, which sounds soft, recur most?

It is what employers most often say graduates lack. In the 2019 survey, 35.9% rated graduates' ability to communicate clearly in writing as 'not very high' (Lait and Lane, 2019, p. 5), and in free-text answers 17 organisations named communication (mostly writing for non-experts), 11 named applying economics to real problems and 13 named data analysis (Lait and Lane, 2019, p. 7). The GES standard expects economists at the SEO grade to present trade-offs and uncertainties clearly to non-economists (Government Economic Service, 2022, p. 15).

Nobody has settled how to communicate uncertainty. Bernanke recommends that the Bank of England scrap its fan charts, on the grounds that their conceptual foundations are weak and the public pays them little attention (Bernanke, 2024, p. 12). The transferable skill is stating your assumptions and what would change your conclusion, not mastering one chart format.

Macro knowledge itself was not the top-rated item. In 2019 only 27.5% of employers rated knowledge of macro variables as very important, down from 39.4% in 2014 to 2015 (Lait and Lane, 2019, p. 5). The sample is mostly public sector and consultancy, so it does not generalise to every employer.

## Tools: which software first?

Spreadsheets first, judging by how these institutions describe their work. Large models are a team's instrument, not one person's starting point.

| Tool | Source |
|:--|:--|
| Spreadsheets and the macro framework | IMF country desks used spreadsheet frameworks more than econometric models (Independent Evaluation Office of the International Monetary Fund, 2014, p. 13); the framework was seen as essential because it gives a consistency check across sectors (Independent Evaluation Office of the International Monetary Fund, 2014, p. 15) |
| R, Stata, SAS, SPSS, Python | GES (Government Economic Service, 2022, p. 12); employers naming data analysis were mostly hiring research economists, econometricians and analysts (Lait and Lane, 2019, p. 7) |
| DSGE, semi-structural, sectoral and VAR models | The model suites at central banks (Bernanke, 2024, p. 16, 17); COMPASS's role at the Bank of England has shrunk considerably (Bernanke, 2024, p. 22); no source asks one person to build them |
| AI and big data | Bernanke thinks AI tools 'seem certain to be increasingly important' (Bernanke, 2024, p. 18); that is the author's view, not a measured result |

## How should a self-study plan be ordered?

Below are four steps I put together, each with the sources behind it. No source tested the order. Figures like 'two pages' and 'one alternative scenario' are mine, picked to keep the task small.

| Step | Output | Sources behind it |
|:--|:--|:--|
| 1. Understand the series | Pick a public macro series; write one page on how it is built, how late it is published and how it gets revised | (Government Economic Service, 2022, p. 10); (Bernanke, 2024, p. 13) |
| 2. Benchmark forecast | Forecast with simple rules (mean, last value, same period last year) and score the error. See [Forecasting demand: when to ask experts, when to build a model](https://staging.khoi.co.uk/en/blog/ban-do-phuong-phap-du-bao) and [Seasonal sales forecasts in Excel](https://staging.khoi.co.uk/en/blog/du-bao-mua-vu-excel-mape) | (Hyndman and Athanasopoulos, 2021); (Bernanke, 2024, p. 9) |
| 3. Forecast log | For each forecast, record the number, the reasons and a review date; score it when the outcome arrives; log each update | (Mellers et al., 2015, p. 274, 277); (Independent Evaluation Office of the International Monetary Fund, 2014, p. 25) |
| 4. Note for a non-specialist | Two pages: conclusion, three assumptions, two risks, one alternative scenario; give it to someone outside the field. See [From GDP to a small firm's margin](https://staging.khoi.co.uk/en/blog/gdp-den-bien-loi-nhuan-homestay) and the framework [Macro-to-micro causal chain](https://staging.khoi.co.uk/en/framework/macro-to-micro-causal-chain) | (Government Economic Service, 2022, p. 15); (Bernanke, 2024, p. 10); (Lait and Lane, 2019, p. 7) |

It suits you if you want to read and use macro analysis for business decisions, or to prepare for an analyst post. It does not suit you if the goal is to build and maintain models at a central bank: the sources place that work in dedicated teams with postgraduate staff (Bernanke, 2024, p. 7, 20), and this plan does not go there.

## Going deeper: what does this evidence not show, and what did I read?

Splitting the map into mindset, skills and tools is this post's frame, not any source's. The GES standard uses three other areas: application of knowledge, analysis of data and effective communication (Government Economic Service, 2022, p. 4). The limits:

- Most sources describe what organisations expect and do not measure whether having the full list produces better analysis. The one measurement about staff is the IMF's experience regression, and it is a correlation; the IMF's own process had also changed substantially in the five years before 2014 (Independent Evaluation Office of the International Monetary Fund, 2014, p. 1, 23).
- Mellers and colleagues state that a strategy winning a tournament is not necessarily optimal; the sample was recruited through professional societies, research centres and word of mouth, all held degrees, 83% were men and 74% US citizens (Mellers et al., 2015, p. 268).
- The survey of 40 employers is about economics graduates in general (Lait and Lane, 2019, p. 2). QAA's 'systems and anticipatory thinking' label comes from sustainable-development guidance, not from research on analysts (Quality Assurance Agency for Higher Education, 2023, p. 6).
- There is no private-sector source and no Vietnamese one. As of 3 November 2026, GOV.UK shows the GES standard last updated on 20 December 2022; the Bernanke review is dated 12 April 2024; the latest employer survey I found is from September 2019.

### What did I read?

GES (2022): the whole 20-page PDF, printed page numbers. Bernanke (2024): pages 1 to 38 and the passages on staffing in Part IV (pages 56 and 57); I did not read the rest of Part III, the other parts of Part IV or the annexes. IEO (2014): the executive summary, paragraphs 34, 35, 37, 42, 43, 79 to 89 and Box 4; I did not read the background studies, so the regression results are the report's own summary; printed page numbers. Mellers and colleagues (2015): the full published text. Lait and Lane (2019): the results page and the report PDF; page 6 of the PDF is a chart that text extraction could not read, so I use no figures from it, and page numbers are PDF page numbers. QAA (2023): all of it. Hyndman and Athanasopoulos (2021): section 5.1 and the start of section 5.2. The evidence-level table, the four-step plan and the remarks marked 'my inference' are my synthesis.

## Frequently asked questions

### Do you need a PhD to work as a macro analyst?

None of the sources I read sets that as a condition. At the Bank of England a little over half of the staff involved in the forecast hold a master's-level qualification and around a third hold a doctorate (Bernanke, 2024, p. 20), but that describes a team, not an entry requirement. The Government Economic Service standard says only a subset of its components applies to any one role and nobody is asked to show experience of every item (Government Economic Service, 2022, p. 5). For roles in Vietnam I have no source.

### Should I learn Excel, R or Python first?

The IMF evaluation's survey of country desk economists found spreadsheet frameworks and judgement used more widely than econometric models (Independent Evaluation Office of the International Monetary Fund, 2014, p. 13). The GES standard names Excel alongside SPSS, Stata, R and SAS (Government Economic Service, 2022, p. 12); of the 13 employers who mentioned data analysis, some named R, Stata and Python and only two mentioned Excel (Lait and Lane, 2019, p. 7). My suggestion, untested: stay in a spreadsheet until you meet a problem it cannot handle, then add one programming language.

### Is it worth copying the superforecasters?

Partly. Mellers and colleagues found that this group beat the others on cognitive ability, the way they updated beliefs and the effort they put in, and concluded that superforecasters are partly discovered and partly created (Mellers et al., 2015, p. 267). But the tournament questions were mainly geopolitical and the participants were recruited forecasters who all held degrees, not a macro data series (Mellers et al., 2015, p. 268).

### What is the evidence for systems thinking and anticipatory thinking?

The only document I found that names them is the QAA benchmark for economics degrees, which lists systems, anticipatory, strategic and critical thinking among core skills and links them to sustainable-development guidance (Quality Assurance Agency for Higher Education, 2023, p. 6). That describes what a degree should teach, not what analysts were measured to use. I rate the evidence low.

## References

- Bernanke, B.S. (2024) *Forecasting for monetary policy making and communication at the Bank of England: a review*. London: Bank of England. Available at: https://www.bankofengland.co.uk/-/media/boe/files/independent-evaluation-office/2024/forecasting-for-monetary-policy-making-and-communication-at-the-bank-of-england-a-review.pdf (Accessed: 2026-11-03).
- Government Economic Service (2022) *GES technical framework 2022*. London: Government Economic Service, HM Government (GOV.UK). Available at: https://www.gov.uk/government/publications/ges-technical-framework-2022 (Accessed: 2026-11-03).
- Hyndman, R.J. and Athanasopoulos, G. (2021) *Forecasting: principles and practice (3rd edn)*. Melbourne: OTexts. Available at: https://otexts.com/fpp3/.
- Independent Evaluation Office of the International Monetary Fund (2014) *IMF forecasts: process, quality, and country perspectives*. Washington, DC: International Monetary Fund. Available at: https://ieo.imf.org/en/-/media/ieo/files/evaluations/completed/03-18-2014-imf-forecasts-process-quality-and-country-perspectives/full-report.pdf (Accessed: 2026-11-03).
- Lait, A. and Lane, S. (2019) *2019 employers survey: survey of employers of economics graduates*. Bristol: The Economics Network, University of Bristol. https://doi.org/10.53593/n2974a
- Mellers, B. et al. (2015) ‘Identifying and cultivating superforecasters as a method of improving probabilistic predictions’, *Perspectives on Psychological Science*, 10(3), pp. 267–281. https://doi.org/10.1177/1745691615577794
- Quality Assurance Agency for Higher Education (2023) *Subject benchmark statement: economics (5th edn)*. Gloucester: Quality Assurance Agency for Higher Education. Available at: https://www.qaa.ac.uk/docs/qaa/sbs/sbs-economics-23.pdf (Accessed: 2026-11-03).

---

Canonical HTML version: https://staging.khoi.co.uk/en/blog/nha-phan-tich-vi-mo-can-gi