Manifesto

I finished an MSc in Economics at Universidad Carlos III de Madrid with a toolkit and a frustration. The toolkit was real: models that say what must be true if their assumptions hold, and methods that separate cause from coincidence. The frustration was that the questions I kept running into outside the classroom, at work, in conversations, in the news and in my own life, were rarely the ones the toolkit was pointed at.

Why does a city keep growing while nobody seems to build? Why does a technology that gets cheaper every month seem to need more people to install it? Who actually ends up better off when a tool makes everyone more productive? These are economic questions. They have economic answers, or at least economic ways of getting closer to one. But they are usually argued about with anecdotes and slogans, or studied years after the fact, when the decisions have already been made.

Project Frontier is my attempt to do the work in between: to take the problems of today seriously as economic problems, while they are still today's.

An independent, personal research project. It is not attached to a university, a think tank, a company or a party, and it has no client. That is a limitation, because I work without a team, a grant or privileged data. It is also a freedom. I can choose questions because they matter, follow them where they lead, and say plainly when the answer is "we don't know yet".

The work comes in three forms. Papers make an argument formally and test it where data allow. Research notes are smaller: a model, a measurement, a bound, an honest attempt that did not fully work. Essays put the same arguments in plain language, for anyone who wants to understand the problem without the algebra.

From noticing things. Many of the questions here started as observations from daily life and work that did not fit the usual explanation: a price that moved when it should not have, a job that grew when it should have shrunk, a gain that did not reach the people who produced it. Economics is useful precisely when intuition and evidence disagree. The frontier in the name is that edge: the place where what we observe has not yet been explained, and where change, technological, demographic or institutional, is happening faster than the explanations.

Two areas run through the work so far. The first is technology, and above all AI: how it changes what firms sell, what workers are paid for and what organisations need to use it. The second is the other big structural problems of the present, starting with housing and migration in Spain.

1. Start from the problem, not the method. A question earns its place by mattering to people's lives, not by fitting a technique.
2. Keep model, calibration and estimation apart. A theorem, a number chosen to illustrate it, and an estimate from data are three different things. Every piece says which one it is offering, and never passes one off as another.
3. Say what is not being claimed. Every paper has a section on what it does not show, and every essay has a box on what would prove it wrong. An argument that cannot be wrong is not worth making.
4. Measure directly when possible. A quantity inferred as a residual is a hypothesis. Wherever the data exist, measure the thing itself.
5. Check the sources. Numbers come from the primary source, not from someone's summary of it. When a figure cannot be traced, it comes out.
6. Publish the data and the code. Anyone should be able to rebuild the results, find the mistakes and do better. Where I build a dataset others can use, I release it.
7. Revise in public. These are working drafts. When a result changes, the new version says what changed and why.
8. Write so that a non-economist can follow. Clarity is not a concession; it is a test of whether the argument is understood.

Not advocacy: the work follows the evidence and reports results that are inconvenient for every side. Not forecasting dressed up as research: when the honest answer is a range or a bound, that is the answer. Not a finished body of knowledge: everything here is provisional and open to criticism, which is welcome.

If you work on these questions, have data that could test an argument here, or think something is wrong, I would like to hear from you. The point of doing this in public is to get closer to the answer with help, not to defend the first version of it.

MSc in Economics, Universidad Carlos III de Madrid (UC3M).