Data Economist

AI Impact Assessment

AI can already perform a significant portion of tasks in this career. Graduates should expect the role to evolve substantially — developing AI-complementary skills will be essential.

Methodology: Anthropic's March 2026 research into real-world AI task adoption across occupations.

Significant Transformation Underway

A degree combining economics and data science remains genuinely valuable in 2026, but you need to be honest about what you are buying. The credential signals quantitative rigour and economic literacy, which employers across finance, government, and consulting still actively seek. The risk is that graduates entering in the next three to five years may find the junior pipeline thinner than it appeared when they enrolled, because AI tools are absorbing the data cleaning and model-running tasks that used to occupy the first two years of a career. The strongest degree programmes will push you into causal inference, experimental design, and policy communication rather than just teaching you to run regressions.

By 2031, AI coding agents and automated data pipelines will handle the bulk of data collection, cleaning, and standard econometric modelling that currently occupies junior data economists. Employers will expect graduates to arrive already comfortable directing these tools rather than performing the underlying tasks manually. Headcount at the entry level will likely contract in financial services and large consultancies, though public sector roles in the ONS, HM Treasury, and local government may prove more stable. Graduates who can operate as 'economist plus AI director' rather than 'economist who also codes' will be best positioned.

By 2036, the data economist role as currently described will have split into two distinct tracks: a smaller, senior track requiring deep causal and policy expertise where human judgement is genuinely irreplaceable, and a broader applied analytics track where the job is increasingly about managing AI-generated outputs and communicating them credibly to non-technical stakeholders. Salaries at the top end should remain strong precisely because the pool of people with genuine economic intuition will not grow as fast as demand for economic insight does. Mid-tier roles, however, face sustained pressure as AI systems become capable of end-to-end analysis on well-defined problems.

By 2046, it is plausible that near-fully automated economic analysis systems handle routine forecasting, policy impact assessment, and market monitoring with minimal human input. The data economist who survives in that landscape will be closer to a scientific director or institutional economist: someone who governs the questions asked, audits model assumptions, and takes responsibility for the societal consequences of data-driven decisions. The profession will be smaller but more prestigious and better paid, similar to how actuarial science contracted and professionalised simultaneously. Students entering today should build towards that senior, judgement-heavy tier from day one rather than expecting a traditional career ladder to carry them there.

Master causal inference, not just correlation

Predictive modelling is the part of this job AI will absorb fastest. Double down on causal inference methods, natural experiments, and quasi-experimental design, because determining whether a policy actually caused an outcome requires the kind of economic reasoning AI still handles poorly. Programmes at UCL, LSE, and Edinburgh that include dedicated causal econometrics modules are worth prioritising for precisely this reason.

Develop a policy communication specialism

The ability to translate complex findings into decisions that policymakers, boards, or the public will actually act on is underrated and undervalued in most data-focused degrees. Seek out placements in government analytical units, parliamentary offices, or think tanks such as the Resolution Foundation or Institute for Fiscal Studies where the output is influence, not just a clean dataset. This positions you in the part of the role that AI consistently underperforms.

Treat AI tools as a core technical competency

You will not out-compute AI systems, so stop competing on that axis. Instead, become expert at directing them: prompt engineering for analytical workflows, evaluating the quality of AI-generated model outputs, and knowing when to override automated conclusions. Graduates who arrive able to do this will be seen as force multipliers rather than candidates whose work AI has just made redundant.

Build sector depth alongside technical breadth

A data economist who deeply understands the energy transition, healthcare funding mechanics, or labour market inequality is far harder to replace than one who is generically skilled at analysis. Choose a sector during your degree through placements, dissertation focus, or extracurricular research, and accumulate genuine domain knowledge that requires years of contextual immersion to develop. Sector expertise compounds in a way that technical skills, increasingly commoditised by AI, no longer do.