Economics, Business & Finance: Plot my own data

What you'll be able to do: Turn your own numbers, spending, hours worked, a small business's sales, into the models you're taught, and find out whether they actually behave the way the textbook says.

Why your own data is the honest test

Textbook curves are clean because they're drawn to teach the shape, not measured from anything. Your own data is noisy, has few observations, and frequently does not look like the textbook curve at all, and that gap is the lesson, not a failure of the exercise. A demand curve fit on eight weeks of your own coffee purchases will not have a clean, confident slope; it will have a scatter of points and a shaky line, and seeing that is worth more than being shown a textbook diagram with the noise already removed. It's also the honest introduction to why real elasticity estimates come with confidence intervals, not single numbers, and why small samples struggle to identify anything precisely (see method 09).

What to plot

A personal demand curve. Track what you actually paid for something you buy repeatedly (coffee, a rideshare, a subscription tier) against how much you bought, across weeks or across venues with different prices. Plot price against quantity and see whether a downward slope shows up in your own noisy data, and roughly how steep it looks.

A personal labour supply curve. If you work an hourly job, or freelance, plot pay rate against hours chosen, across different weeks or gigs at different rates. Ask whether more hours appear at higher pay throughout your range, or whether there's a point where higher pay corresponds to fewer hours, the backward-bending labour supply curve, where above some income level people start choosing more leisure over more pay. Whether your own data shows this bend is genuinely uncertain going in; that uncertainty is the point, not a flaw in the exercise.

A small business's books. Take a real or invented shop's transaction list (sales, purchases, rent, wages) and build the trial balance and income statement yourself before asking anything to check it. The exercise isn't the arithmetic; it's turning a list of events into the formal accounting story of what happened.

A portfolio or a single stock. Plot a real stock's price history, compute its realised volatility over different windows, and compare it to a broad index over the same period. Ask what the chart would need to show for you to conclude the stock is "riskier" than the index, rather than just eyeballing the wiggles.

The exercise

Here's my own data: [paste your numbers, dates, prices, quantities, hours, whatever you tracked]. Turn it into a chart appropriate to what I'm trying to show. Don't interpret it for me yet, first ask me what I predict the shape will be, based on the model we've studied. Then show me the chart and ask me to say, specifically, where it matches the model and where it doesn't.

Predicting first, before seeing the plotted result, is what makes this a test rather than a demonstration, you're checking whether you actually believe the model enough to bet on its shape.

Reading the gap between your data and the textbook

When your own chart doesn't match the clean textbook version, and with a handful of noisy real observations, it usually won't, the interesting question is why, and there are several honest answers:

• Too little data to identify anything. Eight data points can't distinguish a real relationship from noise; this is the sample-size problem every empirical economist works around, and feeling it on your own dataset makes the abstract warning concrete.
• Something else moved at the same time. You bought more coffee on cold weeks regardless of price, a confound, not a refutation of demand theory (this connects directly to method 09, identification).
• The model's scope doesn't cover your case. A subscription price you can't actually vary in response to (you're locked into an annual plan) isn't testing a demand curve at all; there's no price variation to respond to.

Distinguishing these three is the actual skill. "My data didn't show a demand curve, so the theory is wrong" skips past all of them.

Across the disciplines

Microeconomics. Own-price elasticity from your own purchases; a rough production function from a part-time job's output per hour worked.

Macroeconomics. Harder to do personally, but plotting your own spending against a published inflation index over the same months is a personal, concrete version of a real price index, and a good way to see how a population-level average can diverge from any one household's experience.

Accounting and finance. Building statements from a real or simulated ledger; plotting a real investment's returns against a benchmark.

Marketing and entrepreneurship. If you run or have run a small side venture, plot price changes against sales volume over time, the closest thing to a real, personal demand-curve estimate most students will ever have direct data for.

Pitfalls

1. Treating a bad fit as proof the model is wrong. Small, noisy, real samples are supposed to look messy; that's the honest state of most real estimation, not a special failure of your data.
2. Skipping the prediction step. Seeing the chart before committing to a guess turns the exercise into a demonstration you watch rather than a test you run.
3. Not controlling for the obvious confound. If price and season both moved, you haven't isolated a price effect, go back to method 09 before drawing a conclusion.
4. Using too few observations and reading a story into noise anyway. Three or four points make almost any story fit; say so rather than reporting a confident slope from them.
5. The tell: every model you've studied has only ever been shown to you in its clean textbook form, never tested against a single real, messy dataset of your own.

Try this today

Pick one thing you can measure about your own economic life this week, what you paid for something across a few purchases, or hours versus pay at a job. Predict the shape of the relationship before you plot it. Then plot it and say specifically where your data agrees with the textbook model and where it doesn't, and why.