Ben Davies

Lots of people criticize economic models for being unrealistic.

Talking to lots of people who know different things helps us learn.

Everyone has beliefs about how they should behave.

My previous post explained why rational people can prefer like-minded information sources.

If people were Bayesian, then giving them more information would help them learn the truth and reach consensus.

Cheng and Hsiaw (2022) study an agent who wants to learn about a binary state (e.g., whether a vaccine is safe).

Science is hard, and communicating it to a broad audience is even harder.

This is a technical follow-up to a previous post on assortative mixing in networks.

A new version of nberwp, an R package containing data on NBER working papers, is available on CRAN.

Most empirical research relies on hypothesis testing.

Suppose I use the Gale-Shapley (GS) algorithm to find a stable matching between two sets \(P\) and \(R\) of size \(n\).

Suppose I want to learn the value of \(\omega\in\{0,1\}\).

This post presents a simple model of corruption in two-party elections.

Suppose I want to run a pilot study of a mental health support program before rolling it out at scale.

Chapter 9 of Kahneman et al. (2021) discusses how predictions made by humans can be less accurate than predictions made using statistical models.

We can think of research as having two phases: a “creative phase,” during which researchers search for new ideas, and a “working phase,” during which researchers exert effort on an idea.

nberwp, an R package providing information on NBER working papers and their authors, is now available on CRAN.

Linear models cannot be estimated when regressors are perfectly correlated, and their coefficients have large variances when regressors are almost-perfectly correlated.

Many competitive runners aim for negative splits: running the second half of a race faster than the first.

My previous post described Gale and Shapley’s (1962) algorithm for solving the stable matching problem.

Let \(A\) and \(B\) be sets of \(n\) people.

This post analyzes the representation of, and collaboration among, female authors of NBER working papers over the last four decades.

This post shows how monopoly insurance pricing can lead to inefficient risk sharing.

Let \([n]\equiv\{1,2,\ldots,n\}\) be a set of individuals.

Let \(N\) be a network with \(n\) nodes, each of which has a “type” belonging to some set \(T\).

Suppose \(X\) and \(Y\) are random variables with $$\DeclareMathOperator{\E}{E} \DeclareMathOperator{\Cov}{Cov} \DeclareMathOperator{\Var}{Var} \newcommand{\abs}[1]{\lvert#1\rvert} Y=\beta X+u,$$ where \(u\) has zero mean and zero correlation with \(X\).

Pancs (2020) proposes an auction for vaccines in which people can bid on others’ behalf.

Last December I compared strategies for playing white elephant, a game in which people take turns either unwrapping a gift or stealing a previously unwrapped gift.

Earlier this month the kākāpō was elected Bird of the Year for 2020.

Suppose 50 people each have four friends.

Suppose some proportion \(\theta\) of the population engages in a socially undesirable activity—say, evading taxes.

Research is increasingly conducted by teams.

“Relatedness, Complexity and Local Growth,” co-authored with Dave Maré while I worked at Motu, has undergone peer review.

When the COVID-19 pandemic began, the New Zealand government faced uncertainty around the virus’ health and economic consequences.

Last week I finished up at Motu, an economic research institute where I worked for two and half years.

Let \(\newcommand{\N}{\mathbb{N}}\N=\{1,2,\ldots\}\) be the set of positive integers.

Suppose we have data \(\{(x_i,y_i):i\in\{1,2,\ldots,n\}\}\) generated by the process $$y_i=\beta x_i+u_i,$$ where the \(u_i\) are random errors with zero means, equal variances, and zero correlations with the \(x_i\).

This week’s Riddler Classic poses a question about bacteria (paraphrased for brevity): Each bacterium in a colony splits into two copies with probability \(p\) and dies with probability \((1-p)\).

Last year I applied to several economics PhD programs at elite universities and business schools.

Research often draws on multiple fields, each contributing field-specific ideas and techniques to the production of new knowledge.

The COVID-19 pandemic exposes policymakers to several sources of uncertainty: How dangerous is the virus?

Let \(X\), \(Y\) and \(Z\) be random variables.

Greedy Pig is a game used to teach probability and statistics to primary school students.

Suppose I want to generate some random graphs that exhibit community structure.

The seminal model of insurance demand (Arrow, 1963; Mossin, 1968) describes a consumer who chooses the level of coverage \(I^*\) that maximises their expected utility $$\phi(I)=(1-p)u(Y-\pi I)+pu(Y-\pi I-L+I),$$ where \(p\) is the probability of suffering…

Suppose I’m training for an upcoming race.

Later this year, New Zealanders will vote in a referendum on whether to legalise voluntary euthanasia under the conditions specified in the End of Life Choice Bill (hereafter “the Bill”).

Running in pairs (and, more generally, in groups) can be more rewarding than running alone.

Spotify offers two music and podcast streaming services: a free, online-only service, and a paid “Premium” service with extra features like unlimited skips and offline playback.

I am excited to announce that I will be moving to the United States later this year to pursue a PhD in economics at Stanford University.