LO
Economics
Trains you to reason about how people respond to prices, rules and incentives — and to work out whether a number showing that response was caused by the thing everyone is crediting for it.
Take one claim from a news article this month that credits a policy or a technology with a change in some number, and write one page: what the number is being compared against, and what else changed at the same time. One page, dated, kept. That is the artefact this degree is supposed to produce, and almost no graduate finishes with one.
A single number for a whole major would hide the thing that matters: this degree trains several separate competencies, and they are not all moving in the same direction. Automation acts on tasks, so any assessment lives on the occupation pages below — not here.
Written for undergraduate economics degrees in Singapore and China, single-honours and joint. Two things vary enough to change the picture: how much econometrics and coding the programme actually requires, and whether it sits inside a business school or a faculty of arts and sciences — the second decides which employers come to campus more than the curriculum does. Research-track programmes leading to a doctorate run on their own timetable and are not assessed here. Economics is also the degree whose graduates scatter widest, so the directions below are a sample of where they land, not a list of what the degree qualifies you for. It qualifies you for nothing in particular, which is both the complaint and the point.
What this degree actually trains#
Not the course list — the competencies underneath it, and whether each one is worth more or less than it was.
Telling what caused what
Worth more than beforeWorking out what a number is being compared against: what would have happened otherwise, who selected themselves into the group you are looking at, and whether the comparison could have come out the other way at all.
Running the estimate got fast; deciding what the estimate is a comparison of did not. A tool will clean a panel, choose a specification and hand back a chart with a confident coefficient in it, and no part of that supplies the design — who the control group is and why the two groups were alike before. The occupation page for data analysts on this site marks knowing when the data is lying as human-led for a neighbouring reason: the judgement rests on history with that particular data, not on the statistics.
Working out who changes behaviour
Worth more than beforeReading a price, a rule or a target for what people will do once it exists — who will qualify for it who was never meant to, who will stop doing the thing that is now being measured, and which of the effects only shows up a year later.
This is the half of policy reasoning that does not survive summarisation, because it is about what has not happened yet and so is in no document to be summarised. It shows up directly in the jobs this degree reaches. The compliance page here describes the hard half of reading a regulation as knowing which of this company's actual processes the clause lands on; the lending page describes an exception as, by definition, a case the model scored wrong. Both are questions about behaviour rather than about text, and both are what the tools hand back to you once the reading is free.
Knowing what a published number counts
HoldingGoing to the method note before quoting the series: what the unemployment rate counts as employment, who a survey's sample frame leaves out, and what happened to the series in the year somebody changed a definition.
Reading a method note became cheap for anybody who thinks to ask for it, and the number of people who quote a series without asking has not fallen. So this is neither displaced nor newly scarce — it is the floor the rest of the degree stands on, and the reason an economics graduate is useful in a room where everyone has the same chart. The failure it prevents is specific and common: comparing two years of a series whose own publisher states, in writing, that it is not built for comparison across years.
Cleaning the panel and running the estimate
Worth less on its ownMerging the sources, handling the missing years, writing the code that produces the table — everything between having the question and having the coefficient.
Structured input, a known output shape, and an immediate signal for whether it worked — the conditions that make a task cheap to hand over, and the same ones this site's data analyst page uses to mark query writing as automating. Doing it by hand is still how you learn what a dataset can and cannot support, so the first few times are not wasted. It is no longer what anybody pays for, and a graduate whose evidence of ability is a clean regression table is holding up the part that got cheap.
Deriving the result from the assumptions
Worth less on its ownSetting up the optimisation, finding the equilibrium, doing the comparative statics — the exam-shaped half of the theory sequence.
Outside research this half rarely sold on its own. It sold as a screen: being able to do hard abstract work quickly was expensive to fake, so employers used it to sort people. It is now cheap to fake, which is a change in the signal rather than in the skill. What survives is the thing the derivations were training you for — noticing which conclusion is an artefact of an assumption nobody would defend out loud, and saying so before someone builds a policy on it.
Where it can lead#
Several directions, never one. Each says what your training reuses, what graduates typically lack, the real entry conditions, and one thing you can test this term.
Credit and lending decisions
task-level analysis →- What transfers
- Selection and moral hazard are the economics of lending, and you were taught them under those names: who applies when the price is set this way, and what a borrower does once the money has arrived. The occupation page describes the exception — the self-employed applicant whose income is real and does not look it — as the case a model cannot find, because an exception is by definition a case the model scored wrong. Recognising that shape is what this degree drills for three years.
- What graduates typically lack
- Graduates commonly arrive with the theory of credit and none of the product: what a facility letter contains, what security actually is, how a file gets built. That gap costs more than it used to, because gathering the file is the task the occupation page marks as automating, and it was the rung on which the product knowledge used to be picked up. The second thing graduates underestimate is the difference between declining a case in a problem set and declining one to a person who is sitting opposite you.
- Entry reality
- Bank campus hiring in both markets takes economics graduates into credit and relationship roles, and it is still one of the wider doors. Expect a smaller junior intake than the year above you had: the file preparation that used to employ that intake is being removed, and nobody has said what replaces the few hundred files it took to learn the judgement above it. Before signing, ask two things — what your approval limit will be, and when it last moved for someone at your level.
Get hold of one real credit application pack — from an internship, a bank's published checklist for small-business lending, or a family business's own file — and write one page sorting the items into two lists: what a system can verify against records somebody already holds, and what needs a person to ask a question. The second list is the job you would be doing.
Regulation, competition and the compliance function
task-level analysis →- What transfers
- Economics is the discipline regulators use to decide whether a market is working, and the reasoning transfers whole into the second line of a regulated firm: what a rule is trying to prevent, what it will actually cause, and who gained from the version that got written. The occupation page names the hard half of reading a regulation as knowing which of this company's processes a clause lands on — a question about how the work is organised, not about the text.
- What graduates typically lack
- Graduates commonly lack the vocabulary and the filing discipline, and underestimate how much of this work is a record that has to survive an inspector arriving years later. A second gap belongs to this degree specifically: economics rewards you for being interesting about a rule, and this job rewards being boring about it in writing and interesting about it in about one meeting a quarter.
- Entry reality
- Steady hiring, and in Singapore a regulator plus a set of licensed institutions that recruit economics graduates directly. It is less contested than the front office because it carries less prestige among finance students, which is an advantage you can use. Read the occupation page for one fact that has nothing to do with technology: whether the compliance function reports into the business it polices decides more about this job than any tool, and you are allowed to ask about it at interview.
Find one consultation paper or regulatory impact assessment published in your market and write one page on two things the document does not say: which behaviour the rule expects to change, and how a firm could comply by reclassifying something rather than by changing what it does. Take that page to an interview; nobody else will have written one.
Measurement and applied analysis
task-level analysis →- What transfers
- Econometrics, pointed at one company's own data. The occupation page marks two of its tasks as human-led — working out what is really being asked, and knowing when the data is lying — and both are this degree's core: what the number is a comparison of, and what happened to the series the month somebody changed a definition. Economics graduates are also unusual in having been drilled to write down what an estimate does not establish, which is the sentence a business almost never gets handed.
- What graduates typically lack
- Code, at a size the coursework never reaches. Most economics graduates have used a statistics package on a dataset that fits on a laptop and has been cleaned once already; the seat wants SQL against tables nobody documented. The second gap is the direction of the question: coursework starts from a dataset and finds a result, while the job starts from a manager's vague worry and has to decide what would even count as an answer before any data is touched.
- Entry reality
- Analytics teams do recruit economics graduates, and the seat has been moving: the occupation page describes analyst effort shifting from answering questions to owning the metric definitions that let everyone else answer their own. The same page says the honest thing about that shift — the definitions job is one per organisation and smaller than the work it replaced. So treat a first analytics post as two years of getting close to one business's data, which is what makes the rest of this page possible, rather than as a destination.
Pick a number your own university publishes about itself — graduate employment, class size, admissions — and reconstruct how it must have been computed. Write down the three choices whoever built it had to make, and which one would move the number most. That page is an artefact showing the thing an interview cannot test for.
Research and policy analysis, inside an institution
- What transfers
- The whole degree, in the order it was taught: the theory to say what should happen, the identification to find out whether it did, and the writing to say what the estimate does not settle. This is the destination the curriculum is actually designed around, which is why it feels like the default direction and is not one.
- What graduates typically lack
- Graduates commonly lack a finished piece of work that was theirs. A dissertation supervised to a deadline is not the same artefact as a question you chose, failed at, and narrowed until it could be answered — and the second is what a research assistant post is selecting for. The other thing graduates underestimate is how long this route runs before it pays anything: an assistant post and then a doctorate is closer to a decade than to a plan.
- Entry reality
- Central banks, ministries, competition authorities, international organisations and university research groups hire pre-doctoral research assistants, and in Singapore several of them recruit locally and openly. The posts are few, selection leans on a named referee and a writing sample, and a doctorate is the gate for anything past assisting. Note what this site cannot give you here: VOLO has no occupation page for an economist or a policy researcher, so there is no task-level analysis to send you to. That is our gap, not a judgement that this direction is untouched.
Replicate one published table from a paper in a field you care about, using the authors' own posted data, and write one paragraph on what broke. Replication is the cheapest available proof that you can do research, almost no undergraduate has done one, and afterwards a referee can say something specific about you instead of something warm.
What to add outside the classroom#
This is about what graduates commonly lack in practice — not a claim that your school failed to teach it.
Enough code to work with data nobody cleaned: SQL, and one language you can write a script in. It is the single largest gap between what this degree hands you and what three of the four directions above expect on the first day.
One question you chose yourself, carried all the way to a finished answer with the data attached. Coursework hands you the dataset and the deadline; what travels between employers is evidence that you can narrow a question until it can actually be answered.
The habit of opening the method note before quoting the series. It costs ten minutes, it is the most common mistake made by people with your training, and it is one of the few things you can be reliably better at than a fluent tool.
This term#
One or two actions, each producing something you can show someone. Not a reading list.
Take one claim from a news article this month that credits a policy or a technology with a change in some number, and write one page: what the number is being compared against, and what else changed at the same time. One page, dated, kept. That is the artefact this degree is supposed to produce, and almost no graduate finishes with one.
Find the method note for one statistic your own coursework used this term, and check whether its publisher says that series may be compared across years. If it does not and your assignment compared it anyway, write that sentence into the assignment. It costs you a mark and buys a habit that will outlast the module.
Common questions#
Look at where it leads rather than at the degree itself. The three occupations economics reaches on this site do not move together, but the exposed task in each one is the same kind of thing: on the lending page it is gathering the file, on the compliance page it is working the alert queue, on the analytics page it is writing queries. All three are what juniors were hired to do, and in all three the task the page marks as human-led is a judgement about what a number or a rule actually does to behaviour. That judgement is what this degree trains, and it is the part holding. What would genuinely change the answer is not AI but you: economics qualifies you for nothing in particular, so it pays when a second thing is added to it and stays vague when nothing is. Moving to a degree that qualifies you for exactly one job is a different bet, not a safer one.
It was always the deal, and the terms moved. The degree never bought a licence; it bought the presumption that you could be put in front of an unfamiliar problem and be useful within a few months. The change is that the entry tasks which used to absorb a generalist graduate for two years while they learned an industry are the ones the occupation pages here mark as automating — the file, the queue, the queries. So the runway is shorter, and the answer is not to abandon the generality but to add the specific thing during the degree rather than after it. One industry you know properly, or enough code to work with real data, is what the two years used to supply.
Yes, and for a different reason than the one in the syllabus. You will write less of the code and read far more output that you did not produce, which makes the valuable half the design rather than the estimation: what comparison a specification is quietly assuming, and which coefficient is real but is measuring who selected into the sample. This site's data analyst page puts the split in almost the same place — writing queries is marked automating, knowing when the data is lying is marked human-led. Learn the part that decides what the number means, and let the tool do the part that produces it.
Two different questions with two different answers. A doctorate is a gate, not an upgrade: the research direction on this page does not open without one, nothing else on the page requires it, and it pays in autonomy rather than in salary — a reasonable destination and a bad hedge. A taught master's mostly buys a second sitting of campus recruiting, plus the econometrics and coding the analytics direction assumes if your undergraduate programme was the non-quantitative kind. Neither buys judgement. Work out which specific door you are trying to open first, then ask two people who went through it in the last three years whether the qualification is what opened it.
The half that produces the table. Cleaning the panel, choosing a specification and generating the output are cheap now, on the same grounds this site uses to mark query writing as automating on the data analyst page: structured input, a known output shape, and an immediate signal for whether it worked. What is not exposed is the half that decides what the table is a comparison of, and what would have to be true for the result to mean what it appears to mean. The uncomfortable part is that the degree examines you mostly on the first half, so your transcript is evidence about the part that got cheap, and the evidence for the other half is something you have to produce yourself.
Method#
Assessments live on tasks, not on majors. Follow any direction above to its occupation page to see which tasks are changing, how strong the evidence is, and what it does not yet show.