Contributing to the textbook Use Hypothes.is, an amazing tool for annotating the web. Go to Hypothes.is, and “get-started”

    includes:
      in_header: [header_include.html]

<script async defer src="https://hypothes.is/embed.js"></script>

head(as.data.frame(nasa))#> lat long month year cloudhigh cloudlow cloudmid ozone pressure #> 1 36.20000 -113.8 1 1995 26.0 7.5 34.5 304 835 #> 2 33.70435 -113.8 1 1995 20.0 11.5 32.5 304 940 #> 3 31.20870 -113.8 1 1995 16.0 16.5 26.0 298 960 #> 4 28.71304 -113.8 1 1995 13.0 20.5 14.5 276 990 #> 5 26.21739 -113.8 1 1995 7.5 26.0 10.5 274 1000 #> 6 23.72174 -113.8 1 1995 8.0 30.0 9.5 264 1000 #> surftemp temperature #> 1 272.7 272.1 #> 2 279.5 282.2 #> 3 284.7 285.2 #> 4 289.3 290.7 #> 5 292.2 292.7 #> 6 294.1 293.6

Now this can be simplified using the new {{}} syntax: summarise_groups <- function(dataframe, grouping_var, column_name){ dataframe %>% group_by({{grouping_var}}) %>% summarise({{column_name}} := mean({{column_name}}, na.rm = TRUE)) } Much easier and cleaner! You still have to use the := operator instead of = for the column name however. Also, from my understanding, if you want to modify the column names, for instance in this case return "mean_height" instead of height you have to keep using the enquo()–!! syntax.

(1) “How likely do you think it is that this hypothesis will be replicated (on a scale from 0% to 100%)?” (2) “How large do you think the standardized effect size (in terms of Cohen’s d) from the replication will be, relative to that in the original paper (on a scale from −50% to 200%)?”, and (3) “How well do you know this topic? (Not at all; Slightly; Moderately; Very well; Extremely well.)”

0.506 (0.532)

For the 12 studies with an original p < 0.005, 10 (83%) replicated. For the 12 studies with an original p > 0.005, only 1 (8%) replicated. Further work is needed to test if prediction markets outperform predictions based only on the initial p-value, to test if the market also aggregates other information important for reproducibility.

At least one of my un-named research co-authors will heartily agree with this statement.↩︎

ps f

List previous command history: history

It’s worth noting that first line of the script starts with #!. It is a special directive which Unix treats differently.

Happy Git and GitHub for the useR

8 Writing, argumentation, presentation, and (Economic) logic: Being clear and making sense

Prediction in Policy

Suppose the algorithm chooses a tree that splits on education but not on age. Conditional on this tree, the estimated coefficients are consistent. But that does not imply that treatment effects do not also vary by age, as education may well covary with age; on other draws of the data, in fact, the same procedure could have chosen a tree that split on age instead

hese heterogenous treatment effects can be used to assign treatments; Misra and Dubé (2016) illustrate this on the problem of price targeting, applying Bayesian regularized methods to a large-scale experiment where prices were randomly assigned

Chernozhukov, Chetverikov, Demirer, Duflo, Hansen, and Newey (2016) take care of high-dimensional controls in treatment effect estimation by solving two simultaneous prediction problems, one in the outcome and one in the treatment equation.

In particular, a set of papers has already introduced regu-larization into the first stage in a high-dimensional setting, including the LASSO (Belloni, Chen, Chernozhukov, and Hansen 2012) and ridge regression (Carrasco 2012; Hansen and Kozbur 2014

These same techniques applied here result in split-sample instrumental variables (Angrist and Krueger 1995) and “jackknife” instrumental variables

Understood this way, the finite-sample biases in instrumental variables are a consequence of overfitting.

Even when we are interested in a parameter β ˆ, the tool we use to recover that parameter may contain (often implicitly) a prediction component. Take the case of linear instrumental variables understood as a two-stage procedure: first regress x = γ′z + δ on the instrument z, then regress y = β′x + ε on the fitted values x ˆ. The first stage is typically handled as an estimation step. But this is effectively a prediction task: only the predictions x ˆ enter the second stage; the coefficients in the first stage are merely a means to these fitted values.

Prediction in the Service of Estimation

New Data

Zhao and Yu (2006) who establish asymptotic model-selection consistency for the LASSO. Besides assuming that the true model is “sparse”—only a few variables are relevant—they also require the “irrepresentable condition” between observables: loosely put, none of the irrelevant covariates can be even moderately related to the set of relevant ones.

First, it encourages the choice of less complex, but wrong models. Even if the best model uses interactions of number of bathrooms with number of rooms, regularization may lead to a choice of a simpler (but worse) model that uses only number of fireplaces. Second, it can bring with it a cousin of omitted variable bias, where we are typically concerned with correlations between observed variables and unobserved ones. Here, when regular-ization excludes some variables, even a correlation between observed variables and other observed (but excluded) ones can create bias in the estimated coefficients.

97the variables are correlated with each other (say the number of rooms of a house and its square-footage), then such variables are substitutes in predicting house prices. Similar predictions can be produced using very different variables. Which variables are actually chosen depends on the specific finite sample.

Through its regularizer, LASSO produces a sparse prediction function, so that many coefficients are zero and are “not used”—in this example, we find that more than half the variables are unused in each run

One obvious problem that arises in making such inferences is the lack of stan-dard errors on the coefficients. Even when machine-learning predictors produce familiar output like linear functions, forming these standard errors can be more complicated than seems at first glance as they would have to account for the model selection itself. In fact, Leeb and Pötscher (2006, 2008) develop conditions under which it is impossible to obtain (uniformly) consistent estimates of the distribution of model parameters after data-driven selection.

First, econometrics can guide design choices, such as the number of folds or the function class.

These choices about how to represent the features will interact with the regularizer and function class: A linear model can reproduce the log base area per room from log base area and log room number easily, while a regression tree would require many splits to do so.

Ta b l e 2Some Machine Learning Algorithms

Picking the prediction func-tion then involves two steps: The first step is, conditional on a level of complexity, to pick the best in-sample loss-minimizing function.8 The second step is to estimate the optimal level of complexity using empirical tuning (as we saw in cross-validating the depth of the tree).

egularization combines with the observability of predic-tion quality to allow us to fit flexible functional forms and still find generalizable structure.

This procedure works because prediction quality is observable: both predic-tions y ˆ and outcomes y are observed. Contrast this with parameter estimation, where typically we must rely on assumptions about the data-generating process to ensure consistency.

In empirical tuning, we create an out-of-sample experiment inside the original sample.

Performance of Different Algorithms in Predicting House Values

We consider 10,000 randomly selected owner-occupied units from the 2011 metropolitan sample of the American Housing Survey. In addition to the values of each unit, we also include 150 variables that contain information about the unit and its location, such as the number of rooms, the base area, and the census region within the United States. To compare different prediction tech-niques, we evaluate how well each approach predicts (log) unit value on a separate hold-out set of 41,808 units from the same sample. All details on the sample and our empirical exercise can be found in an online appendix available with this paper athttp://e-jep.org

Making sense of complex data such as images and text often involves a prediction pre-processing step.

The fundamental insight behind these breakthroughs is as much statis-tical as computational. Machine intelligence became possible once researchers stopped approaching intelligence tasks procedurally and began tackling them empirically.

Why not also use it to learn something about the “underlying model”: specifically, why not use it to make infer-ences about the underlying data-generating process?

Economic theory and content expertise play a crucial role in guiding where the algorithm looks for structure first. This is the sense in which “simply throw it all in” is an unreasonable way to understand or run these machine learning algo-rithms.

available finite-sample guidance on its implementation—such as heuristics for the number of folds (usually five to ten) or the “one standard-error rule” for tuning the LASSO (Hastie, Tibshirani, and Friedman 2009)—has a more ad-hoc flavor.

Should out-of-sample performance be estimated using some known correction for overfitting (such as an adjusted R2 when it is available) or using cross-validation?

Kernel regressionKernel bandwidth

final category is in direct policy applications. Deciding which teacher to hire implicitly involves a prediction task (what added value will a given teacher have?), one that is intimately tied to the causal question of the value of an additional teacher.

In another category of applications, the key object of interest is actually a parameter β, but the inference procedures (often implicitly) contain a prediction task. For example, the first stage of a linear instrumental variables regres-sion is effectively prediction. The same is true when estimating heterogeneous treatment effects, testing for effects on multiple outcomes in experiments, and flexibly controlling for observed confounders.

Machine Learning: An Applied Econometric Approach

David Chalmers: I think I may have introduced this actually years ago. I call it “The law of minimization of mystery”. I was making fun of some people who wanted to tie consciousness and quantum mechanics. But I do think there are interesting potential links between the two. I mean. The problem in quantum mechanics is not just that it’s mysterious.It’s a very specific problem. How do the standard dynamics of quantum mechanics which has these two processes: Schrodinger evolution and wave function collapse. Mostly the wave function evolves this way, except on certain times, when you make a measurement, the wave function changes in this special way, and that’s totally standard quantum mechanics.

I’ve argued you need something new in the story and the kind of view I’ve been drawn towards are views that take consciousness as something sort of fundamental and irreducible in our picture of the natural world in the same way that we take space and time and mass and charge as fundamental. We’re used to the idea that some things are fundamental. If you can’t explain electromagnetic phenomena in terms of your old laws, your old properties and laws, spacetime, mass, Newtonian dynamics, you bring in something else. Electric charge, Maxwell’s laws.Likewise, I think for consciousness. So I’ve been drawn towards views that take consciousness as fundamental and what that comes to in practice in philosophy is either you’ve got the choice between either a dualist view where you’ve got. You’ve got the physical world, which is one thing, and then you’ve got say the mind, you’ve got consciousness, which is another thing.They’re both fundamental properties distinct from each other. And then there are laws that connect them. That’s one view. And the other view is panpsychism, which says consciousness is somehow present at the very basis of the physical world and maybe the physics that we know and love basically somehow fundamentally involves consciousness somehow.

think the most interesting of which is that this whole idea of consciousness is an illusion. A pathology built up by our cognitive systems to believe we have these special properties of consciousness introspectively, even though we don’t.

David Chalmers: Yeah, but I also think there are these sociological effects where most people think… we got this on the PhilPapers survey that most people think that most people think a certain thing, even though most people think the opposite.

I’m sure that many of them are among listeners potentially. It seems like there’s a bit of a stream of this among rationalists and I often find natural scientists, I just can’t get them to accept that there’s like anything strange about consciousness existing.

David Chalmers: Well, the big obvious thing that can be said in defense of philosophy here is the thing that I said already. Which is philosophy by its nature is the field where there’s disagreement, because once we obtain methods for producing an agreement on questions and reasonably decisive ways, we spin it off and it’s no longer philosophy.So from that perspective, philosophy has been this incredibly effective incubator of disciplines. Physics span out of philosophy. Psychology span out of philosophy. Arguably to some extent, economics and linguistics span out of philosophy. So what usually happens is not that we entirely solve a whole of a philosophical problem, but we come up with some methods of say, making progress experimentally or formally on a certain subquestion or aspects of that question and then that gets spun off. The part that we haven’t figured out how to think about well enough that remains philosophy.

the highest correlation coefficient in this survey was 0.56 which is kind of only moderate and it was between moral realism and cognitivism which obviously have a lot to do with one another directly

Robert Wiblin: Yeah. I guess in politics it seems that that brings out people’s tribal instincts, so they tend to group together for practical reasons, if not intellectual reasons, like kind of all sharing the same views or like wanting to fall into line and are particularly incentivized to do that. An interesting thing, I’ll provide a link to a study looking at how ideologically tightly grouped are people in politics, which found that uneducated people just like have views all over the place. Their views on one question don’t really predict their views on another.

David Chalmers: I don’t know. Where I come from, 0.56 is a pretty high correlation coefficient. So I don’t know. Maybe it depends on the area.

And I think at that point, wherein something like in a verbal dispute, which can happen in these cases where you have different people mean different things by free will and well, I think that diagnosis is more apt for some of these questions than for others, but like this is the case again where your average philosopher uses free will, well the thing tied to moral responsibility. Many people outside philosophy may still think, ah, why use free will for that? I want to use free will for this other thing. This ability to fundamentally go against the laws of nature. And then we just have a difference about which one is worth caring about.

Now someone interested in free will could say, well, that wasn’t what I cared about (moral responsibility).

What is it like to be you right now? You’re seeing this text on the screen, you smell the coffee next to you, feel the warmth of the cup, and hear your housemates arguing about whether Home Alone was better than Home Alone 2: Lost in New York. There’s a lot going on in your head — your conscious experiences

Also see slides 1-7 of the powerpoint HERE

Update: Free online access to this text may be available through your university/library (this works at Exeter) via www.vlebooks.com

1.2 Readings and resources

. (2018: but this won’t be focal on the exam.)

Losing face PDF

Losing face

David Chalmers on the nature and ethics of consciousness

Prior exams with answers and ‘mock exams’

Bee2024 finishing insurance problem from sixth problem set

Shorter 'rooftop' video going through 3 cases of 'two-person Public goods' via powerpoint

Power rule

Asymmetric information (Moral hazard, adverse selection, signaling) Separate notes/handouts, to be integrated in Readings tbd Behavioural economics – first lecture (limits to cognition, willpower, self-interest), second lecture (applications) [11] NS: Ch 17, plus supplements T1:11 Further readings tbd Eighth problem set: Behavioural economics Overall each reading is optional, but to do well on the exam you need to do at least some of these readings, and/or cover some of the more advanced material. Supplementary reading: theory Amos Tversky & Daniel Kahneman, 1979. “Prospect Theory: An Analysis of Decision under Risk” (Seminal) Supplementary reading: applications and empirical work DellaVigna, Stefano. “Psychology and economics: Evidence from the field.” Journal of Economic literature 47.2 (2009): 315-372. Benartzi, S. & Thaler, R.H., 2007. Heuristics and biases in retirement savings behavior. The journal of economic perspectives, pp.81-104. Farber, H., 2008. Reference-dependent preferences and labor supply: The case of New York City taxi drivers. The American Economic Review. Available at: http://www.ingentaconnect.com/content/aea/aer/2008/00000098/00000003/art00021 [Accessed November 19, 2015]. (Behavioural insights team) EAST: Four simple ways to apply behavioural insight Kellner, Reinstein and Riener, 2016. Conditional generosity and uncertain income: Field and lab evidence Material linked at giveifyouwin.org

The payoff from staying Silent (cooperating) in each period is: −2×(1+g+g2+g3+...)−2×(1+g+g2+g3+...)-2 \times (1 + g + g^2 + g^3 + ... ) Here I get -2 in each period, starting today. Discounting this, we add up -2 (today), −2g−2g-2g (next period), −2g2−2g2-2g^2 (the period after next), etc, as represented above. The payoff from Confessing right away (after which both players Confess always) is: −1−3×(g+g2+g3+...)−1−3×(g+g2+g3+...) -1 -3 \times (g + g^2 + g^3 + ... ) Formula for a geometric series (where 0<g<10<g<10<g<1): g+g2+g3+g4...=g/(1−g)g+g2+g3+g4...=g/(1−g)g + g^2 + g^3 + g^4 ... = g/(1-g) Note on Maths: The standard derivation of this, which is pretty neat, is in the text. This formula is an important one in economics (and beyond), particularly for discounting a constant stream of payoffs, e.g., stock dividends Thus cooperation in a single period is ‘weakly preferred’ (at least as good) if (−2)×(1+g+g2+g3+...)≥(−1)+−3×(g+g2+g3+...)(−2)×(1+g+g2+g3+...)≥(−1)+−3×(g+g2+g3+...)(-2) \times (1 + g + g^2 + g^3 + ... ) \geq (-1) + -3 \times (g + g^2 + g^3 + ...) g+g2+g3+...≥1g+g2+g3+...≥1g + g^2 + g^3 + ... \geq 1 Note on the intuition for the second formula: the left side is loss of future payoffs (-3 vs -2 forever from next period, so a loss of 1 per period starting tomorrow). The right side is gain in ‘the present’ period (getting -1 rather than -2), so it is un-discounted. g/(1−g)≥1g/(1−g)≥1g/(1-g) \geq 1 g≥12

Individual optimisation solved problem

Nash proved that if we allow mixed strategies, then every game with a finite number of players in which each player can choose from finitely many pure strategies has at least one Nash equilibrium.

More volatile underlying assets will translate to higher options premiums, because with volatility there is a greater probability that the options will end up in-the-money at expiration.

The option is European and can only be exercised at expiration.No dividends are paid out during the life of the option.Markets are efficient (i.e., market movements cannot be predicted).There are no transaction costs in buying the option.The risk-free rate and volatility of the underlying are known and constant.The returns on the underlying are normally distributed.

In low-income countries the vast majority are unwilling to pay for effective drugs simply because they are unable to pay. Low-income nations need more price discrimination—and vastly lower prices—if they are ever to afford the world's most effective medicines.

She found a German seller offering packs of the same nappies she buys in Luxembourg for the same price she normally pays. Looking more closely at the unit price, however, Nadine realised that the German packs contained 140 nappies, whereas the packs in Luxembourg had only 90, making them much more expensive. She switched straight away to buying all her nappies from the German shop.

Channel Tunnel

map_dbl(df, mean)

out[i] <- fun(df[[i]])

Holt, C., and S. Laury (2002), Risk Aversion and Incentive Effects, American Economic Review, v. 92 (5): 1644-1655. Crosetto, Paolo, and Antonio Filippin. “A theoretical and experimental appraisal of four risk elicitation methods.” Experimental Economics 19, no. 3 (2016): 613-641. Pedroni, Andreas, Renato Frey, Adrian Bruhin, Gilles Dutilh, Ralph Hertwig, and Jörg Rieskamp. “The risk elicitation puzzle.” Nature Human Behaviour 1, no. 11 (2017): 803.

11.11 Who benefits from 3DPD?

University strikes: Students demand tuition fee refunds for planned walkouts

Find the pure-strategy Nash equilibrium or equilibria

3. Now draw the farmers’ ‘best response functions’ in a diagram.

Do parts A and B; part C is optional enrichment

ero marginal cost for one more user.

This second assumption, called diminishing marginal utility, will imply ‘risk aversion’!

Two statistics about reducing your risk of an early death made headlines around the world recently. The first seems to be a great reason to add a four-legged friend to your life. It suggests that owning a dog is tied to lowering your chance of dying early by nearly a quarter. The second statistic claims that even a minimal amount of running is linked to reducing your risk of premature death by up to 30%. Ruth Alexander finds out what’s behind these numbers and we hear from epidemiologist, Gideon Meyerowitz-Katz.

2018 final exam with suggested answer guidelines

Should we help companies tailor prices to your wage packet?

writing

LaTeX is a high-quality system equipped with special features for technical and scientific documentation. A great tool for thesis help due to its user-friendly interface and dozens of helpful features. For example, the tool automatically generates bibliographies and indexes.

NS: Ch 16 – public goods section only (skip Lindahl equiliibrium; sections on median voter and single-peaked preferences are optional)

Ans: B and D

No, because if it were at such a point, it could raise its price and its revenue would increase and costs would decline.

Paying the vertical distance between B and D (sacrificing this amount of ‘other goods’) to obtain the second ticket will leave her at point D. This is on the same indifference curve aas point B so this leaves her equally well off. Thus this is the most she would be willing to pay for the second ticket.

No, because if it were at such a point, it could raise its price and its revenue would increase and costs would decline.

Mock mid-term exam

Individual optimisation solved problem

Individual optimisation solved problem

‘people buy less when the price rises, all else equal’.

MRT(K,L)MRT(K,L)MRT(K,L) equal

MRT(K,L)MRT(K,L)MRT(K,L)

Sugden, Robert. “Looking for a Psychology for the Inner Rational Agent.” Social Theory and Practice 41, no. 4 (2015): 579-98.. ‘Predicting Hunger: The Effects of Appetite and Delay on Choice’; Read and van Leeuwen, 1998 … and related work… worth getting a sense of this, particularly in the context of the Sugden (2015) reading, such as Kahneman and Richard H. Thaler. 2006. “Anomalies: Utility Maximization and Experienced Utility.” Journal of Economic Perspectives, 20 (1): 221-234

Waldfogel, Joel. “The deadweight loss of Christmas.” The American Economic Review 83.5 (1993): 1328-1336. Reinstein, David. “The Economics of the Gift.” (2014; working paper and chapter in Gift Giving and the’embedded’Economy in the Ancient World, F Carlà, M Gori - 2014). D. Colander \Edgeworth’s Hedonimeter and the Quest to Measure Utility’’ Journal of Economic Perspectives, Spring 2007: 215-225. Sugden, Robert. “Looking for a Psychology for the Inner Rational Agent.” Social Theory and Practice 41, no. 4 (2015): 579-98..

Friedman, Milton. “The methodology of positive economics.” (1953): 259. Classic but perhaps outdated Sugden, Robert. “Credible worlds: the status of theoretical models in economics.” Journal of Economic Methodology 7.1 (2000): 1-31. Sugden, Robert; Sitzia, Stefania. “Implementing theoretical models in the laboratory, and what this can and cannot achieve.” Journal of Economic Methodology, Vol. 18, 2011, p. 323-343.

AUT: Lecture 3 - Axioms of Consumer Preference and the Theory of Choice (note, slightly different notation)

Waldfogel, Joel. “The deadweight loss of Christmas.” The American Economic Review 83.5 (1993): 1328-1336. Reinstein, David. “The Economics of the Gift.” (2014). D. Colander \Edgeworth’s Hedonimeter and the Quest to Measure Utility’’ Journal of Economic Perspectives, Spring 2007: 215-225.

Loomes, Graham, Chris Starmer, and Robert Sugden, 1991.“Observing Violations of Transitivity by Experimental Methods”. jstor link Choi, Syngjoo, et al. “Who is (more) rational?.” The American Economic Review 104.6 (2014): 1518-1550.a.

For maths, see also Mathematical methods for economic theory … also free online!

Here Y has a diminishing MU, but for X MU is constant

(Left) when goods are perfect substitutes, there is no Substitution effect of a price change. For prices lll and l′l′l', the minimum-cost consumption that attains U1U1U_1 is the same point (A,B)

common measure of the burden: ‘amount a person would be willing to pay to avoid tax’, for a given revenue raised *A step-by-step illu

2. That is, at this consumption plan he or she would be willing to give up 3 beer

f*h'(r)

Goals of this problem set: Re-acquaintance with mathematical approaches to Economics (e.g., simultaneous equations, graphing functions) Revising the supply and demand model and its implications, applying this to real-world problems, considering empirical approaches Understanding the logic of ‘difficult’ multiple choice questions (assessment tips) Discussing and writing a coherent response to applied Economics questions

Note: The lecturing ‘slides’ are posted contemporaneously to each lecture, but there is no need to look at them – this web-book (‘handout’ is king). OK, fine… the link to the slides is HERE, but don’t look at them. (If you do you will need to press ‘o’ to see the navigation map.

3 Math tools and empirical analysis - – coverage outline

4 Utility and Choice (L2)

2 Economic Models (L1-2)

Alfred Marshall argued that a good’s price must equal both the cost to produce and the value (to consumers) of the last unit produced and consumed.

With ‘single crossing’ (a fancy condition you can ignore in this module) there is a unique price where Qs(p)=Qd(p)Qs(p)=Qd(p)Q^s(p)=Q^d(p), and a unique quantity where the last unit’s value to the consumer equals its cost to produce.

1∗1

GTA’s: Ten minutes, i.e., minutes 5-15 of of tutorial, perhaps (after introducing yourself etc).

Goals of this problem set:

Not ‘output’ – ‘nets provided’ nor ‘paintings purchased’ (Total or marginal) impact per dollar = output per dollar ××\times impact per output

Little giving to the most effective charities ‘Innumerate empathy’: key barrier to promoting effective giving?

When Pandoc converts Markdown to another output format, it uses a template under the hood. The template is a plain-text file that contains some variables of the form $variable$. These variables will be replaced by their values generated by Pandoc. Below is a very brief template for HTML output:

When used locally, this feature requires that reveal.js runs from a local web server.

Vague questions

4.9 …Stating empirical results

“Has the legacy of British institutions increased or decrease the level of GDP in former colonies?”

“how do risk-averse individuals choose among a set of uncertain gambles?

Fields

3.4 Theoretical vs. Applied (focus)

The book is written in RMarkdown with bookdown. It is automatically rebuilt from source by travis. R4DS is a collaborative effort and many people have contributed fixes and improvements via pull request.

e your thesis meaningfu

just

This tool is designed to help you inform your marketing, improve retention of your donors and ultimately increase your income. It means you can target your donors in a really timely way, when you are most at risk of losing them and when they are most likely to donate, maximising your impact whilst still adhering to The Fundraising Regulations.

sample(1:6, 1)

A scatter plot represents two dimensional data