People care about whether their data are shared, but not so much where their data are stored

There has been a substantial policy movement in favour of the localisation of data storage over the past decade (for example, see here). Policymakers often justify data localisation policies by appealing to consumers' supposed preference for having their data stored locally. In particular, they refer to privacy concerns, lack of trust in data handling practices in other countries, and preference for supporting local data storage firms. However, the evidence that consumers have strong preferences for data localisation is very thin. In fact, this new article by Jeffrey Prince (Indiana University) and Scott Wallsten (Technology Policy Institute), published in the journal Information Economics and Policy (ungated earlier version here), may represent the first attempt to really evaluate consumers preferences for data storage.

Prince and Wallsten use a discrete choice survey to evaluate preferences for localisation for different types of data. Specifically:

We constructed five different survey structures, one each centered on the respondent’s smartphone, financial institution, healthcare app, smart home device, and social media. The data types we consider include home address, phone number, income, financial activity, health status and activity, biometrics, music preferences, location, networks, and communications. Across the five survey structures and range of data types, we measure the relative value of full privacy (no data sharing) versus sharing only domestically (localization), sharing domestically and internationally (no localization), and sharing domestically and internationally excluding China and Russia (no localization but with limits). We administered each of these five different surveys across seven different countries: the United States, the United Kingdom, South Korea, Japan, Italy, India, and France

Their sample, drawn from Dynata's online panel, is 11,375 respondents, with 325 completed surveys for each of the five survey structures, for each of the seven countries. Each respondent was shown ten different discrete choice questions. In each question, respondents would have been shown hypothetical alternative scenarios about how their data could be stored and shared, and had to pick their preferred alternative. However, the article doesn't make clear how many alternatives the respondent was choosing from in each choice task, nor whether they simply chose the best of the alternatives, or provided a full ranking of all of the alternatives. Those are issues that are consequential for the analysis, but probably don't bias the results in any way.

In terms of data localisation, Prince and Wallsten distinguish between data not being shared at all, and data being stored and shared domestically only, internationally, or internationally while excluding China and Russia. The latter is included because consumers may be more concerned about their data being stored in China or Russia than being stored in other countries. First, Prince and Wallsten find that:

...virtually all of our parameter estimates are highly significant. As these are estimates of (dis) utility from sharing data in one of three ways (domestically only, internationally, internationally except China and Russia) versus not sharing, the consistent, negative and statistically significant estimates imply that respondents across all of our countries are averse to sharing their data.

In other words, people really don't like their data being shared, regardless of how or where it would be shared. However, in terms of data localisation, Prince and Wallsten find that:

...it is evident that there are a just handful of data types for which we find any notable data localization premium: bank balance, facial recognition, home address, and phone number, all with multiple instances, and voiceprint, with one instance.

Interpreting these results, Prince and Wallsten note that:

...the data types for which we find a data localization premium are also the data types for which citizens find the most value in having no sharing of any kind... citizens across our seven countries, by and large, place little to no value in data localization requirements, despite placing value on full privacy for these data (i.e., no domestic or international sharing)...

In addition, there are no differences between sharing internationally, and sharing internationally while excluding China and Russia. If anything, there is some weak evidence that respondents in South Korea and Japan preferred to have their data shared with China and Russia. That’s striking given how often policymakers highlight the dangers of data flowing to China and Russia. Prince and Wallsten conclude that:

Our findings have several implications. First, they suggest that the use of privacy concerns as motivation for data localization laws may be overstated, although there may be some gross welfare gains for some types of data. Our findings also indicate that if international sharing is allowed, restricting prominent authoritarian countries such as China and Russia appears to have little impact on consumer value, at least for a number of highly populated countries...

...our findings do provide a counterweight to any claim that citizens find value from imposing constraints on international data sharing.

It may still be worthwhile for policymakers to insist on data localisation. Of course, this is just one study (albeit the first study) using survey data from an online panel, so we should be cautious about overgeneralising. Nevertheless, based on this study, the argument that data localisation reflects consumers' preferences for data storage does not hold up to scrutiny. If they want to keep pushing data localisation, policymakers will need to lean on geopolitical or protectionist arguments instead.

Shots fired at the end of a debate on contingent valuation

I have written a number of posts about debates on the contingent valuation method (most recently here, but see the links at the end of this post for more). A 2016 debate that I blogged about here, was picked up again in 2020 (but I didn't blog about it then because I was kind of busy trying to manage the COVID lockdown-online teaching debacle). So, what happened? The first of two 2020 articles published in the journal Ecological Economics (sorry I don't see an ungated version online) is by John Whitehead (Appalachian State University), a serial participant in contingent valuation debates.

This part of the debate centres on 'adding up tests', which essentially test for scope problems. To reiterate (from this post):

Scope problems arise when you think about a good that is made up of component parts. If you ask people how much they are willing to pay for Good A and how much they are willing to pay for Good B, the sum of those two WTP values often turns out to be much more than what people would tell you they are willing to pay for Good A and Good B together. This issue is one I encountered early in my research career, in joint work with Ian Bateman and Andreas Tsoumas (ungated earlier version here).

An 'adding up test' tests for whether the willingness to pay for the global good (Good A and Good B together) is more than adding the willingness to pay for Good A alone to the willingness-to-pay for Good B alone. In relation to this particular debate, Whitehead summarises where we are up to:

Desvousges et al. (2012) reinterpret the two-scenario scope test in Chapman et al. (2009) as a three-scenario adding-up test. They then assert that the implicit third willingness-to-pay estimate is not of adequate size. Whitehead (2016) critiques the notion of the adding-up test as an adequacy test and proposes a measure to assess the economic significance of the scope test: scope elasticity. Chapman et al. (2016) argue that Desvousges et al. (2012) misinterpret their scope test. Desvousges et al. (2016) reply that they did not misinterpret the Chapman et al. (2009) scope test and assert that their adding-up test in Desvousges et al. (2015) demonstrates one of their points.

Desvousges et al. (2015) field the Chapman et al. (2009) survey with new sample data collected with a different survey sample mode than that used by Chapman et al. (2009) and three additional scenarios. Desvousges et al. (2015) conduct an adding-up test and argue that willingness-to-pay (WTP) for the whole should be equal to willingness-to-pay for the sum of four parts (the first, second, third and fourth increment scenarios). Desvousges et al. (2015) find that “The sum of the four increments … is about three times as large as the value of the whole” (p. 566).

Whitehead joins the debate on the side of Chapman et al., defending them by examining Desvousges et al.'s analysis and showing that it actually does meet an 'adding up test', thereby showing that there are no scope problems in the original Chapman et al. paper. Whitehead concludes that there are a number of problems in the Desvousges et al. analysis:

First, they do not elicit WTP estimates explicitly consistent with the theory of the adding-up test. Their survey design suggests that a one-tailed test be conducted where the sum of the WTP parts is expected to be greater than the WTP whole. Second, there are several data quality problems: non-monotonicity, flat portions over wide ranges of the bid function and fat tails. Each of these data problems leads to high variability in mean WTP across estimation approach and larger standard errors than those associated with nonparametric estimators that rely on smoothed data.

I'm not going to get into the weeds here, because what I want to highlight is the response by William Desvousges, Kristy Mathews (both independent consultants), and Kenneth Train (University of California - Berkeley), also published in the journal Ecological Economics (and also no ungated version available). The response is only two pages long, and is a very effective takedown of Whitehead. Along the way, Desvousges et al. note that Whitehead:

...made numerous mistakes in his calculations... When these errors are corrected, adding-up fails for each theoretically valid parametric model that Whitehead used.

One example of Whitehead's errors is:

He used medians for the tests instead of means, assuming – incorrectly – that the sum of medians is the median of the sum.

That's a fair criticism. However, Desvousges et al. are not satisfied leaving it at that. Instead, they go onto the attack:

Also, we examined the papers authored or co-authored by Whitehead that are cited in the recent reviews... These papers provide 15 CV datasets. Each of the three problems that Whitehead identified for our paper is evidenced in these datasets:

• Non-monotonicity: 12 of the 15 datasets exhibit non-monotonicity.
• Flat portions of the response curve: All 15 datasets have flat areas for at least half of the possible adjacent prompts, and 4 datasets have flat areas for all adjacent prompts.
• Fat tails: In our data, the yes-share at the highest cost prompt ranged from 15 to 45%, depending on the program increment. In Whitehead's studies, the share ranged from 14 to 53%.

If Whitehead's data are no worse than typical CV studies, then his papers indicate the pervasiveness of these problems in CV studies.

Ouch! That seems to have ended that particular debate. My takeaway (apart from not messing with Desvousges et al.) is that the contingent valuation method is far from perfect. In particular, it is vulnerable to scope problems (which my own research with Ian Bateman and Andreas Tsoumas (ungated earlier version here) showed some years ago. Ironically, that contingent valuation has particular problems is a message that John Whitehead himself has also argued (see here).

Read more:

• Nothing good happens after midnight when using the CVM
• The ongoing contingent valuation debate
• More contingent valuation debates
• Yet another contingent valuation debate

In wildfires, people prefer to save people rather than endangered species

If you were an incident controller who needed to deploy firefighting resources in a wildfire, how would you decide where to distribute those resources? If there is not enough firefighting to cover all areas at once, which areas should receive priority? Saving human lives seems like it should be a priority, but what about animal lives? What about preserving biodiversity, or saving endangered species from the fire? What about built infrastructure? What about important cultural artifacts? Some of these questions may seem easy to resolve, but there are important trade-offs, and understanding those trade-offs is important.

That is where this 2024 article by John Woinarski, Stephen Garnett, and Kerstin Zander (all Charles Darwin University), published in the journal Conservation Biology (open access, with non-technical summary on The Conversation), comes in. They surveyed a sample of over 2000 Australians, asking them to repeatedly make best-worst choices among five different alternatives (of eleven total). As they explain:

...respondents are asked to state which item among a set of items they consider as best and worst... In our survey, best meant the asset the respondent most wanted to save and worst meant the asset the respondent least wanted to save.

By getting the research participants to repeat this task many times (eleven times, in fact), with different sets of items to choose from, Woinarski et al. develop a good picture of the relative ranking of each of the eleven items, both for each research participant and for the sample overall. This best-worst scaling (BWS) method is a form of non-market valuation, since it essentially works out the relative value (in terms of ranking) of the different options that research participants are presented with. [*]

The eleven options that research participants were ranking overall were:

1. A person with a car stuck behind a fallen tree, whom you know had not received advice to evacuate;
2. A person with car stuck behind a fallen tree, whom you know had ignored repeated advice to evacuate beforehand;
3. A house that you know has no people in it;
4. A farm shed with some hay bales and a tractor;
5. A flock of 50 sheep—a few of which will be killed by fire, but survivors are likely to be badly injured;
6. A population of 50 koalas—a few of which will be killed by fire, but survivors are likely to be badly injured;
7. The last population of a native snail species for which the fire will kill all individuals, thereby causing the species’ extinction;
8. The last population of a small native shrub, for which the fire will kill all plants, thereby causing the species’ extinction;
9. One of only two populations of a rare wallaby for which the fire will kill all individuals of one of the populations (but not affect the other), thereby making it more endangered;
10. Ancient rock art that will be destroyed if fire gets into the weeds now growing in the rock shelter; and
11. An old tree with an ancient Aboriginal carving on the trunk.

The results are interesting, if not terribly surprising:

In terms of relative importance, saving a person who ignored evacuation advice was rated 57% as important as saving a person who had not received warnings... Saving the koala population was rated slightly lower (56% as important as saving a person who had not received warnings). Saving the wallaby population was 45% as important as saving a person who was not warned. Saving the house and shed had the lowest rankings (14% and 9%, respectively, as important as saving a person who was not warned).

For completeness, compared with saving a person who had not received warnings, saving the shrub was rated as 25% as important. Saving the sheep was rated as 26% as important, saving the snails was rated as 25% as important, saving the ancient rock art was rated as 15% as important, and saving the carved tree was rated 12% as important, respectively. Woinarski et al. bemoan that no one loves snails, but I also think the loss of the cultural artifacts would be a tragedy as well. I guess that reflects that each of us would place different weightings on things, and come out with different rankings. And that is what Woinarski et al. look at next, finding that:

Female respondents placed higher importance than male respondents on the protection of the rare wallaby population, the koala population, the sheep, and the tree carving and lower importance than male respondents on the protection of the house, shed, native shrub, and rock art... Older respondents (>65 years) rated protecting people more highly than younger respondents, but rated the tree carving less highly than younger respondents...

Respondents who self-identified as Indigenous placed a higher score on protecting the rock art and tree carvings than those identifying as non-Indigenous.

Those differences may not come as a surprise either. Now, in my ECONS102 class, when we discuss non-market valuation (specifically in the context of estimating the value of a statistical life), I point out that personal experience of the risk makes a difference. And that is true in this case as well. Woinarski et al. find that:

Survey respondents affected by wildfires and those assessing themselves as being prepared for wildfires were less likely to save a person who had not received warnings... Those who rated themselves as prepared for wildfire were also less likely to save a person who ignored warnings, whereas those who had been affected by wildfire were more likely to do so.

It is interesting to consider what the differences mean here. If a person has personal experience of wildfires, then they know how devastating they can be, and how unpredictable and fast-moving. In my mind, that should make them more likely to want to save a person who has not received warnings, but instead they are less likely. On the other hand, it does make sense that they would be more likely to save someone who ignored warnings. Woinarski et al. don't provide a good explanation for that result (although, to be fair, they are focused on the results related to conservation, rather than humans!). On the other hand, people who are well prepared being less willing to help those who ignored warnings makes some sense.

The takeaway message from this paper, though, is that people prefer to save people, rather than endangered species. Especially snails.

*****

[*] If one of the options had been monetary, Woinarski et al. could have used their results to work out the rough monetary value of each option.

What is the most valuable superpower?

How much would you be willing to pay to have superhero powers? Obviously, the answer depends on the type of superhero powers, so let me be more specific. How much would you be willing to pay to be able to fly? Or have mind control? Or teleport? Or to have superhuman strength? These are the questions that this recent article by Julian Hwang (West Virginia University) and Dongso Lee (Korea Rural Economic Institute), published in the Journal of Cultural Economics (ungated version here) attempts to answer.

Hwang and Lee conducted a discrete choice experiment, which involved asking research participants to choose a superpower. However, each alternative came with a 'price' measured in terms of a shorter life expectancy. So, Hwang and Lee note that the resulting estimate of willingness-to-pay is really a 'willingness-to-sacrifice', since the cost is expected years of life foregone.

Their sample is made up of 51 undergraduates at the University of Florida. Each research participant was presented with ten choice tasks, each of which looked something like this (from Figure 1 in the paper):

Hwang and Lee then use a mixed logit model to estimate the willingness to sacrifice (WTS) for each of the four superpowers, for two experimental groups. The treatment group was asked to swear that they will give truthful answers to each question, while the control group was not. Hwang and Lee find that, for the treatment group:

...the mean WTS for mind-control, flight, teleportation, and supernatural physical strength is 3.2 years, 2.07 years, 5.04 years, and 2.95 years, respectively. For the control group, the mean WTS is 2.04 years, 2.95 years, 4.01 years, and 3.9 years, respectively.

Hwang and Lee then use the value of a statistical life-year to estimate the willingness-to-pay for each superpower, finding that:

The mean WTP for mind-control, flight, teleportation, and supernatural physical strength is $332,579, $215,137, $523,812, and $306,596, respectively.

So it appears that, of the four superpowers that Hwang and Lee asked about, teleportation is viewed as the most valuable. However, to a large extent, the results depend on how each superpower is described to the research participants. For teleportation, research participants were told:

You can transport a person or object from one point to another without traveling the physical space between them

You can also transport yourself

You can visit any places you want without spending money or time

The other superpowers were somewhat more limited. Mind-control was limited to controlling a single person, for five minutes at a time. Flight was limited to 100 miles per day. Super-strength was Captain America strength (the ability to press 800 pounds), not Superman-level strength. In comparison, the teleportation power does seem fairly unconstrained, so it's little wonder that it was valued the highest.

This study could definitely be built on, in at least two ways. First, if a study focused on a single superpower (flight, for example), it should be possible to recover the willingness-to-sacrifice for different aspects of the superpower - duration, maximum height, maximum flight speed, whether the superhero needs to remain awake in order to fly, and so on. Second, it would be interesting to know if there is a difference in the willingness-to-sacrifice between comic book fans (or superhero fans more generally) and other people.

These sorts of follow-up questions might even make a good project for a suitably interested (and motivated) Honours or Masters student. And before you think that the subject matter is not important, it is really the ability to apply the tools of non-market valuation (and discrete choice modelling) that is the important aspect of those sorts of projects. As well as just being a fun research question to think about!

[HT: Marginal Revolution, last year]

How much is your data worth to Google?

Our data is valuable to Google. Google's business model is based on using what Google knows about us to target advertisements. The more data that Google has about us, and the more accurate that data, the more precisely they can target advertising, and the more an advertiser will be willing to pay for that advertising.

We currently give our data to Google for free. It is a condition of using free services like Google Search, GMail, Google Maps, and so on. However, imagine for a moment that, instead of giving our data to Google for nothing, Google had to pay us for our data. How much do you think Google would be willing to pay each year for your data? Would Google be willing to pay $1000 per year? $10,000? $50,000? Google is incredibly profitable. Surely our data must be incredibly valuable.

In a post last month on the Asymmetric Information substack, Dave Heatley provides an indication of the answer to how valuable our data are to Google:

Last week, a Google witness in the US v. Google anti-trust trial accidentally blurted out that Apple gets a 36% cut of Google’s ad revenue from being the default search engine in Apple’s Safari browser... This was the final piece of data I needed to scrape together an estimate of the average search revenue that Google receives from each Apple user... As Apple users are, on average, richer than non-Apple users, we can treat this as an upper bound on the value to Google of an individual search user...

How many Apple users are there?

As of February 2023, Apple had more than 2 billion active devices. While iPhones predominate, this total also includes iPads, Macs and Apple Watches. This corresponds to 1.4bn individual users...

What does Google pay to advertise to them?

Google reportedly paid US$18 billion a year to be Apple’s default search engine in 2021. More recent estimates put this at around US$20bn.

That means Google is paying Apple just US$10 per year per device, or US$14 per individual Apple user, to be able to show search-related advertising.

The big reveal: just how valuable is your data to Google?

And how much revenue does Google make from that “investment”? In the ongoing US v. Google trial, University of Chicago professor Kevin Murphy, appearing for Google, revealed that Google pays Apple 36% of the revenue it earns through search advertising.

Using that figure, Google’s average annual revenue is US$28 per Apple device, or US$39 per Apple user. After paying fees to Apple, net revenue is at most 64% of that — that is, US$18 per device or US$25 per user. (The actual figure will be smaller, as it costs Google to operate their platform for both searchers and advertisers.)

So, there you have it. Your personal data is worth no more than US$25 (NZ$43) per year to Google.

If Google was required to purchase access to our data, Google would be willing to pay no more than US$25 per year to do so. If you wanted to hold out for a better deal, Google would simply say, "no thanks", and you'd lose access to their services.

So, if the value of our data to Google is just $US25 per year, why are Google's revenue and profits so high? The answer is scale. If Google has four billion regular users, each providing Google with data worth US$25 per year, that is up to US$100 billion of revenue per year (and in fact, Google's advertising business generates even more revenue than that). Even though each user's data is not worth much, the sheer number of users means high revenue and profits for Google.

Now think about the other side of our bargain with Google. How much value does Google actually give us in exchange for our data? I'd suggest that it is much more than US$25 per year. Think about all the Google services that you use. If you had to pay real money for them (and for any alternative to them), how much would you be willing to pay? We don't have an answer to that question, but we do know that people would be willing to pay between US$600 and US$2000 for access to Facebook (see here and here), which they also receive for nothing. I'd suggest that Google provides much more value than Facebook does, and so we'd be willing to pay more for Google services than for Facebook.

Taken altogether, we give our data to Google, and that data is worth US$25 per year to Google. Google provides us with services that are worth far more than US$25 per year, probably in the high hundreds of dollars or more per year. If we marketised both sides of this exchange, we could charge Google US$25 per year for our data, but Google could charge hundreds of dollars or more per year for their services.

We may gripe about Google (and Meta, and other big tech firms), and there are definitely negative aspects to their activities (more on that in my book review tomorrow). However, when you actually look at it, the big tech firms generally do appear to be giving us a good deal.

Using virtual reality to improve discrete choice experiments

Following on from my post earlier this month about eye-tracking in discrete choice experiments, I recently read this new article by Ilias Mokas (Hasselt University) and co-authors, published in the Journal of Environmental Economics and Management (open access). They look at the impact of using virtual reality in a discrete choice experiment. To see why, let's first take a step back.

Discrete choice experiments involve giving research participants a series of choices over hypothetical 'goods'. Essentially, research participants are asked how much they are willing to pay for the hypothetical good. It might be an environmental good (clearer air; clearer water; reforestation, etc.). Or it might be road safety improvements, or a hypothetical vaccine. The key is that there is no market for the good now, so it is difficult to value. But knowing how people value the good is important for decision-making, such as how to allocate scarce public resources.

The problem with a discrete choice experiment is that the good is hypothetical. A 'standard' discrete choice experiment simply describes the options in words. Research participants can't see the options they are asking to choose between. It is therefore difficult for them to evaluate how much they are willing to pay for each option, and than manifests in uncertainty for the research participants, and ultimately in a large component of randomness to the estimated willingness-to-pay (WTP) in discrete choice experiments.

In the Mokas et al. paper, the choices were about the amount and style of 'green infrastructure', essentially how many and how big the trees are along the side of the street. That is a context where a textual description isn't really going to do it justice. You could show the research participants some pictures of a rendered hypothetical streetscape, or a fly-through video. At least those options would ensure that research participants are valuing the same thing, rather than what might be very different interpretations of the textual description. However, Mokas et al. take this a step further and reason that a virtual reality (VR) streetscape would be even better, because it enables the research participants to interact with the environment, moving around, looking at what interests them most, etc. Here's what the set-up looks like (from Figure 1 of the article):

Mokas et al. test the effect of VR with 180 Belgian research participants, and a streetscape that is representative of the Flanders region. Comparing three treatments (text, a fly-through video, and VR), they find that:

...the choice set representation with an immersive VR environment reduced the respondents’ uncertainty, and thus, improved the evaluability for the urban green options compared to a representation with video on a computer screen... A potential explanation for this observation is that while in the text version, the participants have to construct the scenarios in their mind, a representation with video or VR enhances familiarity with the attributes in the choice sets because of the additional visual information provided, which allows to process and evaluate the information more systemically.

In other words, the randomness in the estimated WTP (part of which arises from the research participants' uncertainty) is reduced by using VR. That leads to 'better' (in the sense of being more precisely estimated) measures of willingness-to-pay. It would also hopefully lead to better decision-making from policy-makers presented with an analysis of how much taxpayers are willing to pay for green infrastructure. There are a lot of similar contexts where using VR would improve the resulting WTP estimates. This is an exciting improvement over the underlying 'standard' approach.

Eye tracking and attribute non-attendance in choice experiments

Some years ago, when I was relatively new to the area of discrete choice experiments (DCEs), I had an interesting discussion with my colleague (and DCE guru) Ric Scarpa. In DCEs, research participants are presented with a series of choices between (usually hypothetical) products (for an example from my own research, with Ric Scarpa and others, see here). Each product is made up of a number of characteristics (e.g. price, quality, colour, etc.) that are referred to as attributes.

Anyway, my conversation with Ric was about what is referred to as 'attribute non-attendance'. That is where, when presented with a choice between two (or more) alternatives, a research participant ignores some of the attributes when making their choice. Attribute non-attendance is problematic, because the conventional models for analysing DCEs assume that research participants are paying attention to all of the attributes. Methods had evolved to deal with this issue, but often required debriefing each research participant and simply asking them if there are any attributes they didn't pay attention to. Of course, research participants often don't know this, or aren't aware of it, or in some choices they pay attention to some attributes, but in other choices they pay attention to different attributes.

In my conversation with Ric, I was trying to argue that we could use eye-tracking technology to uncover more systematically research participants' attribute non-attendance behaviour. My theory was that, if a research participant spends more time looking at a particular attribute, then they are giving it more careful thought and attention, and that would indicate that there is no non-attendance towards that attribute. Anyway it turns out that, as with many of my cool research ideas over the years, someone else had already thought of it. Sure enough - research papers using eye-tracking to control for attribute non-attendance started to appear in the research literature soon after that.

However, there is another aspect to eye-tracking which I didn't consider, and which is covered in this 2017 article by Kelvin Balcombe (University of Reading) and co-authors, published in the Journal of Economic Behavior and Organization (ungated earlier version here). Balcombe et al. report on a DCE designed to identify people's preferences for country-of-origin labelling of meat in the U.K. (and the particular context they used was labelling of pepperoni pizzas). However, their research questions also made use of eye-tracking:

Question 1: Do ‘higher value’ (more attractive) attributes attract more visual attention (ceteris paribus)?
Question 2: If individuals or groups have higher visual attention (relative to other attributes) can we infer that those individuals or groups value that attribute more highly than other attributes?
Question 3: If one individual has higher visual attention towards a particular attribute (relative to other individuals) can we infer that this individual values that attribute more highly than other individuals?

They measured visual attention in terms of 'dwell time' - that is, how long each research participant spent looking at each attribute. Based on a sample of 100 research participants (which seems small, but remember that each participant is making lots of choices, in this case 24 choices each), they find that:

...there is, overall, a reasonable correspondence between ET data and other measures of attribute use such as the frequently employed debriefing questions that have become widely reported in the literature. But, our results confirm once again, that at the individual level stated attendance is a very weak signal in relation to visual attendance and vice versa... we find evidence of longer engagement with high value attributes, as measured by total dwell time as well as total number of fixations. This relationship exists, but it is quite weak. This result bolsters existing work that suggests that ET data does reveal something about how respondents value the attributes used in a specific DCE.

So, it appears that eye-tracking data doesn't only reveal information about attribute non-attendance, but may also be used to extract additional information about how much the research participants value each of the attributes (or, at least, to provide more precise estimates of the value of each attribute). However, as Balcombe et al. note, using eye-tracking is cumbersome and expensive. Is the extra cost and trouble worthwhile? They conclude no:

...if the purpose of generating ET data is to improve the efficiency of estimation then we would recommend increasing sample size as a better strategy to pursue.

In other words, you'd be better off putting your scarce resource funds towards recruiting more research participants, rather than adding eye-tracking technology to the battery of tools you employ. Anyway, it was interesting to see where this research idea has gone (the paper gives a good update on the state of the literature, at least as it stood in 2017). It's a pity I wasn't five or ten years earlier in thinking of it! 

Yet another contingent valuation debate

There is something about contingent valuation as a methodology that seems to generate seemingly endless debates in the research literature (see here and here, for example). The contingent valuation method is a type of non-market valuation - a way of valuing goods (and services) that are not (and cannot be) traded in markets. For example, it can be used to value environmental goods (e.g. how much is improved water quality in a stream worth?) or more general public goods (e.g. how much is another bridge across the Waikato River worth?). Essentially, contingent valuation involves presenting research participants with a series of hypothetical choices in order to determine how much they would be willing to pay for the good or service (it is what we refer to as a stated preference method). For instance, you might use a survey that asks whether people would be willing to pay an additional $5 in council property taxes to improve water quality in a particular stream (or $10, or $50, or $500), and then use that data to work out how much (on average) people are willing to pay for improved water quality.

One of the challenges in contingent valuation is whether the payment is expressed as a single one-off payment (e.g. pay $100 once and the water quality will improve) or as a series of regular (e.g. annual) payments (e.g. pay $25 per year and the water quality will improve). The latest debate I read recently addresses this important question.

First, this 2015 article by Kevin Egan (University of Toledo), Jay Corrigan (Kenyon College), and Daryl Dwyer (University of Toledo), published in the Journal of Environmental Economics and Management (sorry, I don't see an ungated version online). Egan et al. argue that there are three reasons that annual payments should be preferred rather than one-off payments, for three reasons:

First, survey respondents are spared from performing complicated present value calculations. When respondents compare their known annual WTP... to a proposed annual payment, discount rates cancel out in their benefit-cost analysis. When asked to make a one-time payment for a long-lasting environmental improvement, respondents must know their personal discount rate and perform the relevant present value calculation, while having complete fungibility of their income across time and no binding budget or liquidity constraints... Second, we conduct a convergent validity test by comparing CV estimates from surveys with one-time and ongoing annual payments to annual consumer surplus estimates from a travel cost analysis. We demonstrate that CV estimates from surveys with ongoing annual payments better match annual travel cost consumer surplus estimates. Third, a behavioral argument based on mental accounting... suggests that survey respondents who mentally set aside a fixed annual dollar amount for charitable giving will feel more constrained by a large one-time payment compared to a relatively small annual payment.

In other words, annual payments work better because they are easier for respondents to interpret, they are more consistent with revealed preference results based on actual behaviour, and they are consistent with a mental accounting story that notes that a one-off payment has an outsize effect on people's reported willingness-to-pay (WTP), compared with smaller annual payments. They go on to support their results with data from a contingent valuation survey of 967 people, investigating their willingness to pay for water quality improvements at Maumee Bay State Park in Ohio.

Enter the second article, by John Whitehead (Appalachian State University), also published in the Journal of Environmental Economics and Management, but in 2018 (ungated earlier version here). Whitehead attacks the original article on the basis of the convergent validity tests. Specifically, looking at the proportion of survey respondents who were willing to pay at different price levels, he notes that as the price increases a smaller proportion should be willing to pay that amount (because demand curves are downward sloping), but that isn't always the case. Although Egan et al. employ a method that allows them to deal with this issue (by merging parts of the sample where the downward sloping demand curve assumption would be violated), Whitehead shows that an alternative method demonstrates that both one-time payments and annual payments have convergent validity, and he concludes that there is no reason therefore to prefer annual payments.

In the third article, published in the same issue of the Journal of Environmental Economics and Management (ungated earlier version here), Egan et al. respond. They extend their earlier analysis and demonstrate convergent validity using different estimators and a variety of sensitivity analyses. However, in order to demonstrate this validity, they first merge data on two different types of annual payments: a perpetual payment (one that would go on forever) and an annual payment for a fixed period of ten years. 

Finally, Whitehead follows up with a further re-analysis in this article, published in Econ Journal Watch (in 2017, weirdly before the previous two articles appeared officially in print - I guess JEEM had had enough of the debate and declined to publish this rejoinder). He concludes that:

Whenever bids are pooled using dichotomous choice data, the researcher implicitly acknowledges that something went wrong with the execution of the study or with the contingent valuation method itself. It might be that (a) bid subsamples are too low to generate enough power to conduct the statistical test, (b) bids are poorly designed (too close together, too far apart, or there is inadequate coverage of the range of WTP), or (c) contingent valuation method respondents are highly inconsistent... Data cleaning and pooling bids should not be considered a valid research method when the research goal is conducting validity tests over payment schedules or any other issue in the contingent valuation method.

Whitehead's criticism is quite valid. Manipulation of data to enable an analysis may be fine when a lower-bound (or upper-bound) estimate is all that is required. When we are testing the validity of a particular method, the standard of proof is somewhat higher. However, along the way the debate lost sight of two things. First, Egan et al. merged two different annual payment types in their broader analysis. They had earlier demonstrated in the convergent validity tests that perpetual annual payments resulted in similar WTP estimates as the travel cost method. To me, they really showed that perpetual payments should be preferred over a fixed number of annual payments, and over one-off payments. When they merged the two annual payment types together, that nuance was lost. Second, convergent validity was only one of the three reasons that Egan et al. originally proposed for why annual payments should be preferred. The debate focused on only one of the reasons, but the other two remain valid.

Overall, my takeaway is that more follow-up research is definitely required on the convergent validity question, but in the meantime it is likely that we should prefer perpetual annual payments in contingent valuation surveys. Whitehead argues that we should use both a one-time payment and annual payments. Given the potential for survey respondent fatigue and confusion, I wouldn't favour that approach. Contingent valuation solves a real problem - valuing environmental and other goods that are not traded in markets. However, we need to recognise that, like any method, it has its limitations.

Read more:

• Nothing good happens after midnight when using the CVM
• The ongoing contingent valuation debate
• More contingent valuation debates

The economic value of thoughts and prayers

The increasingly clichéd response to any tragedy is to offer 'thoughts and prayers' to the victims and their loved ones. The cynical among us note that this is done in order to make the well-wisher feel better. But, is the offer of thoughts and prayers valued by the recipients? That is the question that this recent article by Linda Thunström (University of Wyoming) and Shiri Noy (Denison University), published in the journal Proceedings of the National Academy of Sciences, sought to address.

Now, obviously, thoughts and prayers are not traded in markets (yet!). So, there is no market price for these services. That means that a non-market valuation techniques is required. Thunström and Noy used an experiment to determine the value that people placed on receiving thoughts and prayers, which could be positive or negative:

Participants were told that a stranger would receive their description and offer a gesture of support in response. We applied a between-subjects study design and Christians and nonreligious participants were randomized into 1 of 4 conditions (C1 to C4). They participated in a [willingness-to-pay]-elicitation mechanism where they could exchange some or all of their $5 for supportive thoughts from a Christian stranger (C1), thoughts from an atheist stranger (C2), prayers from a Christian stranger (C3), or prayers from a priest (C4).

Essentially, the participants could give up some of the $5 they received for participating in the experiment, in exchange for thoughts and prayers from others. The results were interesting. They found that:

...on average, Christians value prayers from a priest at $7.17... and prayers from a Christian stranger at $4.36... In contrast, the nonreligious are “prayer averse”: on average, they are willing to pay $3.54... for a Christian stranger not to pray for them... Likewise, they are willing to pay a priest $1.66... not to pray for them...

So, thoughts and prayers have positive value for Christians (but interestingly, only from other Christians and not from an atheist), but have negative value for the non-religious (which included Atheists and Agnostics). In some further analysis, they found that Christians were more likely to agree with statements about the helpfulness of thoughts and prayers. That suggests that Christians are willing to pay for thoughts and prayers because they expect those thoughts and prayers to convey benefits on them.

That reminded me of this 2010 article (open access) by Nobel Prize winner James Heckman, published in the journal Economic Inquiry. The article is entitled "The effect of prayer on God’s attitude toward mankind", and Heckman concludes that:

A little prayer does no good and may make things worse. Much prayer helps a lot.

Presumably, that also demonstrates the benefits of prayer. [*]

[HT for the Thunström and Noy article: Elizabeth Oldfield in Unherd, via Marginal Revolution]

*****

[*] Actually, the article by Heckman isn't serious. He was using the analysis of prayer to illustrate the foolishness of this earlier article by R.S. Singh, published in the Journal of the Royal Statistical Society in 1977.

Are people willing to pay to avoid harm from international honey laundering?

You probably had to read the title of this post a couple of times. Yes, it does say honey laundering, with an "h". It's taken from the title of this paper, by Chian Jones Ritten (University of Wyoming) and co-authors, published in the Australian Journal of Agricultural and Resource Economics earlier this year (sorry, I don't see an ungated version). The paper focuses on food fraud, which refers to "the intentional substitution, addition, tampering or misrepresentation of food, food ingredients, or food packaging". They note that:

Evidence suggests that Chinese honey is being transshipped and relabelled to mask the true origin of the honey to avoid large tariffs and potential bans, also known as honey laundering... The practice of honey laundering is so prolific that an estimated one-third of honey available for sale in the United States is illegally imported from China and may contain illegal antibiotics and heavy metals...

The focus on Chinese honey is important because:

Chinese honey has the potential to contain illegal and unsafe antibiotics (specifically, Chloramphenicol, enrofloxacin, and ciprofloxacin) and high levels of herbicides and pesticides...

But do honey consumers care? Essentially, Jones Ritten et al. ran an experiment to test whether consumers were willing to pay a US$2.48 premium for an eight-ounce jar of locally produced honey, and tested whether consumers who were first given information about "the negative health implications of honey laundering" were more willing to pay the premium. They found that:

In total, 53.38% of participants across the treatments chose local honey at a $2.48 premium over honey of unknown origin...

Once we control for only honey preferences and use, access to honey laundering information significantly increases the probability (P < 0.10) of participants being willing to pay a $2.48 premium for an 8-ounce jar of local honey... When also including the influence of demographic variables on the probability of paying the premium, honey laundering information still significantly increases the probability (P < 0.05) of participants choosing local honey...

The results suggest that providing honey laundering information increases the probability of participants being willing to pay the premium by as much as 27 percentage points...

So, consumer information does affect consumers' stated preferences for honey, but not for everyone. However, my last sentence also highlights the problem with this study. They only asked what the consumers would do (their stated preference), and didn't actually require a honey purchase (which would be a revealed preference). So, we don't know whether the consumers would actually follow through on their stated preference. Perhaps they could have required a honey purchase?

It also turns out that older consumers were more willing to pay the premium, which led the authors to conclude that:

Targeting older consumers will most likely be successful at garnering more local consumers that are willing to pay for local honey than targeting younger consumers.

However, if older consumers are already more willing to buy locally produced honey, that doesn't mean that the information had a bigger effect on them. The authors could have tested that directly with their data, by running separate analyses for different age groups, or interacting the experimental treatment variable with age, but for some reason they chose not to.

Aside from those issues, it is a nice study. If you want to avoid international honey laundering, buy local.

What is Facebook worth to you?

Last year, I wrote a post about a research paper by Erik Brynjolfsson (MIT) et al., which estimated the consumer surplus generated by Facebook. The authors used non-market valuation techniques - essentially, they asked people how much money they would accept to give up Facebook for a month. The average was $48.49 in 2016, which decreased to $37.76 in 2017.

That was just one estimate though, so before we accept it, it should be replicable. I just read this 2018 paper, by Jay Corrigan (Kenyon College), Saleem Alhabash (Michigan State University), Matthew Rousu (Susquehanna University), and Sean Cash (Tufts University), published in the open access journal PLoS ONE. They use four different samples to estimate what people would be willing to aceept to give up Facebook for differing time periods, between one hour and one year. For three of the samples, they are able to enforce the choice (otherwise, the participant wouldn't be paid). Since consumer surplus is the difference between what a person would be willing to pay for the good or service (or what they would be willing to accept to give it up), and what they actually pay (which in this case is zero), the willingness-to-accept is a measure of consumer surplus.

Converting the estimates from the various samples, with different timeframes, to the equivalent value for a year, Corrigan et al. find the following:

The results are reasonably consistent across the different samples. The first three columns are scaled up from the willingness-to-accept to give up Facebook for one day, three days, and one week respectively, while the last three columns are willingness-to-accept to give up Facebook for a year, for three different samples (a student sample, a community sample, and an online sample). The results are much larger than those of Brynjolfsson (MIT) et al., whose estimates imply an annual value of less than US$600. Clearly, there is more to do in this space.

The estimate of the value provided by Facebook isn't just a fanciful exercise. Goods and services that are provided to consumers for 'free', such as Facebook and other social networks, search engines, video sites like YouTube, and blogs, as well as the increasing phenomenon of 'free' online games like Fortnite, present a bit of a problem for estimating aggregate economic output. The traditional measure of aggregate output, GDP, measures the total value of all goods and services produced, based on their market prices. If the market price is zero, then there is zero contribution to GDP. And yet, these services obviously generate substantial value for society (otherwise, people wouldn't be willing to accept such a high value before they would give them up). Corrigan et al. estimate that:

...across all three samples, the mean bid to deactivate Facebook for a year exceeded $1,000. Even the most conservative of these mean [Willingness-to-Accept] estimates, if applied to Facebook’s 214 million U.S. users, suggests an annual value of over $240 billion to users.

We already know that GDP is not a good measure of societal wellbeing. However, the increasing prevalence of free goods and services is making GDP even worse as a measure over time. We need to find an alternative measure of wellbeing, but it isn't clear at this point what such a measure would be.

[HT: Marginal Revolution, late last year]

Read more:

• How much money would you accept to give up Facebook for a month?

What happens when you disconnect from Facebook?

I've written a few posts on whether Facebook or internet use makes you unhappy (see here and here and here). The problem with most (if not all) earlier studies is that they show a negative correlation between Facebook use and happiness (or life satisfaction), but fail to show a causal relationship. It might be that unhappier people are more likely to use Facebook, or to more intensively use Facebook, than happier people. Or maybe there is some third factor (e.g. work satisfaction) that affects both Facebook use (more satisfied workers use Facebook less) and happiness (more satisfied workers are happier).

A new working paper by Hunt Allcott (New York University) and co-authors (recently covered by the New York Times) addresses this by using a randomised controlled trial - they randomly selected some of their 2844 research participants to switch off Facebook for four weeks, while others only switched off Facebook for one day. They then looked at the effects of that period on a battery of different measures of online and offline activity, news knowledge, political knowledge and views, and life satisfaction, based on a comparison of the treatment group (those that switched off Facebook for four weeks) and the control group (those that switched off Facebook for a single day). As is increasingly common, they had a pre-registered analysis plan, which limits the degrees of freedom to manipulate the analysis to achieve a preferred statistical result. So the results are fairly believable.

Allcott et al. found that:

Deactivating Facebook freed up 60 minutes per day for the average person in our Treatment group. The Treatment group actually spent less time on both non-Facebook social media and other online activities, while devoting more time to a range of offline activities such as watching television alone and spending time with friends and family. The Treatment group did not change its consumption of any other online or offline news sources and reported spending 15 percent less time consuming news...

The fact that Facebook use declined is not surprising, but other online activities also declined, showing that Facebook and other online activities (including online news consumption) are complements, rather than substitutes. Moving on, they also found that:

Consistent with the reported reduction in news consumption, we find that Facebook deactivation significantly reduced news knowledge and attention to politics. The Treatment group was less likely to say they follow news about politics or the President, and less able to correctly answer factual questions about recent news events. Our overall index of news knowledge fell by 0.19 standard deviations. There is no detectable effect on political engagement, as measured by voter turnout in the midterm election and the likelihood of clicking on email links to support political causes. Deactivation significantly reduced polarization of views on policy issues and a measure of exposure to polarizing news. Deactivation did not statistically significantly reduce affective polarization (i.e. negative feelings about the other political party) or polarization in factual beliefs about current events, although the coefficient estimates also point in that direction. Our overall index of political polarization fell by 0.16 standard deviations...

We might decry Facebook as a source of fake news, but it appears to also be a significant source of real news knowledge as well, as shown by the decrease in political knowledge from deactivating Facebook. To be clear, this result arises mainly because it makes people less sure about the news statements they were presented with in the survey (and asked if the statements were true, or false, or if they were unsure). As many would expect though, it appears that Facebook contributes to political polarization. Finally, in terms of happiness or life satisfaction:

Deactivation caused small but significant improvements in well-being, and in particular on self-reported happiness, life satisfaction, depression, and anxiety. Effects on subjective well-being as measured by responses to brief daily text messages are positive but not significant. Our overall index of subjective well-being improved by 0.09 standard deviations... These results are consistent with prior studies suggesting that Facebook may have adverse effects on mental health.

Interestingly, these outcomes were about two-thirds smaller than the effects measured in past correlational studies (which they demonstrate in the paper). So perhaps Facebook isn't as negative for our overall wellbeing as it has been portrayed. However, it is worth noting that the participants that deactivated Facebook were also more likely to reduce their Facebook use after the experiment concluded. Allcott et al.'s results are also:

...consistent with reverse causality, for example if people who are lonely or depressed spending more time on Facebook, or with omitted variables, for example if lower socio-economic status is associated with both heavy use and lower well-being.

Finally, their data allows them to estimate the consumer surplus of Facebook, which is essentially a measure of the total benefits generated by Facebook for consumers. [*] This is because they asked people how much they were willing to accept to deactivate Facebook for a month - a form of non-market valuation (they are not the first to do this, as I noted in this post last year). They estimate this consumer surplus for US consumers at US$230 billion to $365 billion per year. So despite the impacts on wellbeing, Facebook does generate a lot of value.

[HT: Marginal Revolution]

Read more:

• How much money would you accept to give up Facebook for a month?
• Does the Internet make people happier?
• The more you use Facebook, the worse you feel
• Quitting Facebook might make you happier

*****

[*] Strictly speaking, the consumer surplus is the amount that consumers would be willing to pay for the service, minus the amount that they actually pay. In this case, consumers don't pay anything for Facebook use (at least monetarily - we voluntarily give Facebook lots of our data, which may or may not be valuable to us!).

Return migrants are willing to accept lower wages in exchange for better institutional quality

Last year, I wrote a post about how return migrants to Vietnam prefer areas with higher-quality institutions. The post was based on the research of one of my PhD students, Ngoc Tran, myself, and Jacques Poot. Ngoc recently submitted her PhD thesis for examination, which is a great achievement. Along the way, she completed four research papers (see here, here, here, and here), and also has a forthcoming book chapter. In this post though, I want to focus on the fourth research paper, as it is the most novel.

In the paper, co-authored between Ngoc Tran, Jacques Poot and I, we looked at the intensity of migrants' preferences to high-quality institutions back in their home country. In other words, we asked how much things like absence of corruption, political stability, and the rule of law, mattered for migrants' decision-making about potentially returning home. To measure the intensity of preferences, we used a novel application of the contingent valuation method.

To do this, we first recognised that there are compensating differentials for working in different locations - in areas with higher-quality amenities (such as higher-quality institutions), wages will be lower than in areas with lower-quality amenities (such as lower-quality institutions). We can exploit this to work out what higher-quality institutions are worth, by looking at how much a migrant's income would have to change to make them indifferent between the area with lower-quality institutions and the area with higher-quality institutions.

We did this by asking Vietnamese migrants in New Zealand two questions:

1. Given your perceptions of the difference in institutional quality between New Zealand and Viet Nam, what would be the smallest level of weekly income before tax in Viet Nam where you would be happy moving back to Viet Nam permanently?; and
2. Now imagine that the institutional quality in Viet Nam changed so that it was equal to New Zealand in all ways (and everything else remained the same). If this happened, what would be the smallest level of weekly income before tax in Viet Nam where you would be happy moving back to Viet Nam permanently?

The first question allowed us to estimate the compensating different based on the current differences in institutional quality and other amenities between the two countries, as well as migration costs. The second question modifies the institutional quality in Vietnam so that it is equal to that in New Zealand, holding everything else (including migration costs) constant. The difference between the answer to Question 1 and the answer to Question 2 provides an estimate of the willingness of the migrant to accept lower wages in exchange for improved institutional quality in Vietnam.

Our estimates show that willingness to pay for an incremental unit improvement in institutional quality in Viet Nam is, on average, NZD 79.80 per week (approximately 33 percent of the average weekly wage in Viet Nam for the same period). Moreover, older migrants are willing to pay more, as are migrants who perceive institutional quality to be more important for the repatriation intentions.

As far as we know, this is the first paper ever to use contingent valuation to measure the intensity of preference for institutional quality, certainly among migrants if not among any population group. Notwithstanding the continuing debate on the use of the contingent valuation method (which I've written about here, here, and here), this was a really innovative piece of work.

Congratulations again to Ngoc on submitting her PhD thesis!

Read more:

• Return migrants to Vietnam prefer areas with higher-quality institutions

Radiation and house prices after the Fukushima nuclear disaster

How much less would you be willing to pay for a house in an area affected by radiation significantly above background levels, compared with an otherwise-identical house that is unaffected by radiation? It's not a crazy question. In the U.S., hundreds of millions of people (including the populations of 26 of the 100 most populous cities) live within 50 miles of a nuclear reactor. Worldwide, there are 21 nuclear plants that each have more than one million people living within 30 kilometres of them.

Of course, nuclear accidents are thankfully rare. But the risk is not zero, and when an accident does occur, as happened in Fukushima in 2011, people can be understandably reluctant to live in the affected areas due to the risks to their health and wellbeing. Obviously, that has a flow-on impact on house prices, even outside the most heavily affected areas.

Hedonic demand theory (or hedonic pricing), which we discussed in my ECONS102 class last week, recognises that when you buy some (or most?) goods you aren't so much buying a single item but really a bundle of characteristics, and each of those characteristics has value. The value of the whole product is the sum of the value of the characteristics that make it up. For example, when you buy a house, you are buying its characteristics (number of bedrooms, number of bathrooms, floor area, land area, location, etc.). When you buy land, you are buying land area, soil quality, slope, location and access to amenities, etc. You are also buying the exposure to current levels of radiation, as well as the risk of future exposures to radiation in the event of a nuclear accident. If each of those characteristics can be separately valued, then you can place a value on how much people are willing to pay to avoid radiation (or alternatively, how much they are willing to accept to live in a radiation-affected area).

In a new paper in the Journal of Regional Science (sorry I don't see an ungated version anywhere online), Alistair Munro (National Graduate Institute for Policy Studies, Japan) looks at the impact of the Fukushima disaster on house prices in Fukushima and Miyagi prefectures, using data from 2009 to 2017. Fukushima prefecture was most affected by radiation as well as the tsunami that led to the nuclear disaster, while neighbouring Miyagi prefecture was only affected by the tsunami. So, differences between the two in terms of changes in house prices can be attributed to differences in radiation levels (once you control for other characteristics of the properties, of course). He finds that:

...across the subsample of noncondominium residence types a 1 percent rise in radiation leads to a 0.051 percent drop in values, while for condominiums treated separately the elasticity is also 0.051. For housing land the elasticity is 0.044, and 0.032 for land with existing buildings if the age of the building is controlled for.

In other words, areas more affected by radiation have lower house prices. How much lower? Munro reports that:

...using a variety of methods... the impact of radiation translates into a one to two million Yen (US$10,000–20,000)... reduction in housing prices for average residential properties.

That is quite substantial, but is not terribly surprising. However, the next part of the paper is very cool. Having established how much less people are willing to pay for living in an area with more radiation, Munro then uses that information plus information on the risk of cancer arising from environmental radiation, to estimate the value of a statistical life (or VSL).

As I will discuss with my ECONS102 class later this semester, the VSL can be estimated by taking the willingness-to-pay for a small reduction in risk of death, and extrapolating that to estimate the willingness-to-pay for a 100% reduction in the risk of death, which can be interpreted as the implicit value of a life. Munro estimates VSL to be in the region of US$4.5-6.4 million, which is similar to VSL estimated in other studies (and other risk contexts). An additional take-away from that analysis is that there isn't a particularly high element of dread associated with avoiding death from radiation (otherwise, people would be willing to pay more to avoid it, and the estimated VSL would be much higher).

Next we really need to know whether the Fukushima disaster affected people's perceptions of nuclear risk in other areas that are near nuclear plants but which weren't affected by the disaster. That would be much more difficult to establish, but potentially much more interesting.

Cost-benefit analysis and cost-effectiveness analysis are different

In yesterday's New Zealand Herald, Jamie McKay wrote about an interview with Environment Minister David Parker. This bit caught my eye (emphasis mine):

DP: Huh! The industry has been consulted for over a decade here! In terms of cost-benefit you don't actually do an analysis on whether you should have clean rivers, that's a value judgement, and the vast majority of New Zealanders think we should have rivers clean enough to swim in. What you use cost-benefit analysis for is to look at what is the most cost effective way of getting there.

No, that's NOT what you use cost-benefit analysis for. At least, it is not helpful to conflate cost-benefit analysis with cost-effectiveness analysis in this way.

Cost-effectiveness analysis evaluates the cost per unit of benefits for some option, where the benefits need not be measured in dollars (e.g. a reduction in nutrients in a stream). Cost-effectiveness analysis is useful when there is more than one way of obtaining benefits, because it tells you that whichever option achieves a unit of benefits at the lowest total cost is the more cost-effective approach.

Cost-benefit analysis is related, but different. It compares the costs of some activity with its benefits, where both the costs and benefits are measured in dollars (so that they are comparable). The outcome of a cost-benefit analysis is technically a measure of cost-effectiveness - it is a measure of the ratio of benefits to costs (in other words, it measures the value of benefits for each dollar of cost, which is the inverse of cost-effectiveness). If this ratio is greater than one, then the benefits of the activity are greater than the costs. If this ratio is less than one, then the benefits are less than the costs. Simple. Cost-benefit analysis is useful if you want to determine whether or not to undertake some action, or to choose between mutually exclusive alternatives.

Cost-benefit analysis for a single option (e.g. for cleaning up a stream) doesn't tell you what is cost-effective, because that is not its purpose. You need to be comparing multiple options to evaluate cost-effectiveness. In the case of clean streams, there probably are several options for clean-up to choose from. Of course, if you conducted multiple cost-benefit analyses for the different options, then you could argue that the option with the highest ratio of benefits to costs is most cost-effective. But cost-benefit analysis would likely be overkill for this purpose.

Cost-effectiveness analysis is easier to conduct than cost-benefit analysis, because for cost-effectiveness analysis you don't need to measure the value of the benefits (in dollars), which can be difficult as it requires non-market valuations of the benefits. Cost-benefit analysis will only be necessary if you have multiple benefits and you want to know the combined benefit (since converting everything to dollar values is a handy way to combine benefits in a single measure). But that is probably not the case for streams, where you can measure the benefits in terms of something like reduced nutrient loads, and converting the benefits to dollar terms would only add an additional source of error to the analysis.

Cost-effectiveness analysis is much more flexible than cost-benefit analysis if, as Parker implies, you've already made the decision to have clean rivers. If cleaning up streams is your sole goal (e.g. based on a measure of a single nutrient load or an index of several nutrient loads), then cost-effectiveness analysis is most likely what you would use to determine the most cost-effective way of getting there, NOT cost-benefit analysis.

How much money would you accept to give up Facebook for a month?

If I offered you $10, would you give up Facebook for a month? What about $20? $50? $150? On the other hand, if you've already bought into #deleteFacebook, then I wouldn't need to offer you anything. Asking how much I would have to pay you to give up Facebook seems like a fanciful question, but it has an important implication.

In economics, consumer surplus is the difference between the maximum that a consumer is willing to pay for a good or service, and what they actually pay for it (the price). You can think of consumer surplus as the 'profit' (or net benefit) that consumers get from buying. In order to measure the total consumer surplus in a market, we need to measure the area between the demand curve (which shows consumers' willingness-to-pay for the good or service) and the price, for the quantity that is purchased in total. So, in order to measure consumer surplus, you need to know the demand curve for the product. In practice, observing different prices and the quantities that consumers buy at those different prices gives us an idea of the shape of the demand curve (leaving aside the identification problem for now).

But what if you have a good or service that is given away for free? How do you estimate the consumer surplus then? That's where the questions in the title and first paragraph of this post come in. And this is pretty much what some researchers did recently, as described in this new NBER Working Paper (ungated version here) by Erik Brynjolfsson (MIT), Felix Eggers (University of Groningen), and Avinash Gannamaneni (MIT). The authors use a specific type of non-market valuation called discrete choice experiments (which I have used in research before, including in this paper):

Specifically, we ask consumers to make a choice between keeping a digital good or taking a monetary equivalent compensation when foregoing it. This approach measures willingness-to-accept rather than willingness-to-pay money and experimentally varies the offered monetary values.

Which is more-or-less the same as the questions I started this post with (although in their experiment, each person was only asked the question in relation to a single monetary value). The interesting thing about their experiment is that it isn't just hypothetical:

In some of the experiments, we enforce the consumers’ choices, for instance be requiring them to give up Facebook for a given period before they get any payment. This makes their choices incentive-compatible: the rational thing to do is tell the truth when comparing alternatives options or being asked about valuations.

Yes, in order to get the money, some consumers (randomly selected) actually had to give up Facebook. Their sample was in the thousands, and they found a median willingness-to-accept (in exchange for giving up Facebook for a month) of $48.49 in 2016, which decreased to $37.76 in 2017. Looking at this willingness-to-accept, it has plausible relationships with demographic and other variables:

The usage of Facebook per week (self-reported, measured on a 5-point scale from “less than 1 hour” to “more than 14 hours”) is a significant predictor for the value of Facebook (p = 0.006). The more time a consumer spends on Facebook, the more likely they are to keep their access... Similarly, the more friends someone has on Facebook (self-reported, measured on a 6-point scale from “less than 50” to “more than 1000”) the more compensation they require to leave Facebook (p = 0.024). In terms of activities on Facebook (measured on a 6-point scale ranging from “never” to “several times a day,”) consumers perceive significantly more value in Facebook the more they post status updates or share pictures and videos (p = 0.010), the more they like and comment (p = 0.018), and play games (p = 0.025). Watching videos is marginally significant (p = 0.080), while using the messenger and chat is associated with no additional value (p = 0.100). Consistently, we find significant substitution effects due other social media services, i.e., Instagram (p = 0.025), and video platforms, i.e., YouTube (p = 0.003). Thus, consumers who also use Instagram or YouTube are more likely to give up Facebook...

...we see that female respondents are more likely to keep Facebook than male users (p = 0.011). The same holds for older consumers (p < 0.001).

The paper goes on to estimate willingness-to-accept values for other digital goods, which imply an annual consumer surplus that is as high as $17,350 for search engines (compared with just $322 for social networks collectively, including Facebook). The confidence intervals on these estimates are quite large (which is just as well - would it really take over $17,000 to get the median person to give up search engines for a year?).

The paper is written from the perspective that consumer surplus is a better measure of welfare than Gross Domestic Product (GDP). This is because, among other issues, when consumers substitute physical goods for digital goods, this reduces GDP even though it increases consumer welfare. If you're interested in thinking about better measures of national wellbeing than GDP, this is an important argument (although reading this paper is probably not the best place to start if you are interested in that - try here instead). If we wanted to better measure national wellbeing, then measuring consumer surplus in all markets (including markets where there is no price) would be a good option. But then we have to find a way to measure consumer surplus in markets where there is no price, and the Brynjolfsson et al. paper shows us one way to do this.

[HT: Marginal Revolution]

Reducing excess demand at the Great Barrier Reef

Late last year, I wrote a post about excess demand for New Zealand's Great Walks:

When a good or service has no monetary cost, there will almost always be excess demand for it - more consumers wanting to take advantage of the service than there is capacity to provide the service. Excess demand can be managed in various ways - one way is to raise the price (as suggested by Sanson). Another is to limit the quantity and use some form of waiting list (as is practiced in the health sector). A third alternative is to degrade the quality of the service until demand matches supply (because as the quality of the service degrades, fewer people will want to avail themselves of it).

The Great Walks are not the only tourist attractions that are subject to excess demand. As Michael Vardon (ANU) wrote recently in The Conversation, the Great Barrier Reef is another example:

The Great Barrier Reef is one of the world’s finest natural wonders. It’s also extraordinarily cheap to visit – perhaps too cheap.

While a visit to the reef can be part of an expensive holiday, the daily fee to enter the Great Barrier Reef Marine Park itself is a measly A$6.50. In contrast, earlier this year I was lucky enough to visit Rwanda’s mountain gorillas and paid a US$750 fee, and the charge has since been doubled to US$1,500...

I understand that some people instinctively object to the idea of trying to put monetary values on things like the Great Barrier Reef. But I think valuation helps, on balance, because it offers a way to assimilate environmental information into the economic processes through which most decisions are made. Money makes the world go around, after all.

However this should be done on the proviso that the valuation is systematic and based on sound environmental and economic data.

Vardon's article is mostly about environmental accounting (and is worth reading if you want to learn a little more about non-market valuation of natural resources). That is, it is about placing a value on the Great Barrier Reef to justify a higher visitor fee. However, it isn't necessary to estimate the Reef's value in order to reduce the tourist pressure on it. If you are worried about there being too many visitors, you simply need to raise the visitor fee. Higher prices reduce excess demand. It is really as simple as that, and if we want to protect these natural resources (Great Walks, Great Barrier Reef, or other natural resources with names that don't start with Great), then higher prices are a simple and reasonably effective way to do so.

What to do about students buying essays?

I recently read this 2015 paper by Dan Rigby (University of Manchester), Michael Burton (University of Western Australia), Kelvin Balcombe (University of Reading), Ian Bateman (University of East Anglia), and Abay Mulatu (London Metropolitan Business School), published in the Journal of Economic Behavior & Organization (ungated here). I thought this was an interesting paper because it applied non-market valuation techniques to a good that is actually sold in markets - essays.

The authors use a specific non-market valuation technique that is called discrete choice modelling (which my colleague Riccardo Scarpa is a world-leading expert in, and which I have also been involved in for a couple of projects, including this one). Discrete choice modelling involves presenting the survey participants (in this case, 90 humanities and science students from three UK universities) with a number of hypothetical choices. Each choice involves a number of goods with different attributes (in this case, the attributes included the price of the essay, the quality of the essay in terms of the grade it would receive, the risk of being caught, and the penalty if caught), and often there is also the choice to buy nothing at all. The participants make several choices, which allows us to determine the implicit weighting the participants place on the different values of the attributes.

In analysing the data, Rigby et al. use a latent class model. I won't go into the detail underlying this, but essentially it determines how many different types of decision-makers there are, with each type placing different weight on the attributes of the good (in this case, essays). They found that there were two types of students, corresponding to students who were very reluctant to buy essays, and those who were more willing to do so. They also found that:

...half of our subjects indicate a willingness to buy one or more essays in the hypothetical essay choice experiment. Students’ stated willingness to participate in the essay market, and their implicit valuation of purchased essays, vary with the characteristics of student and institutional environment. Risk preferring students, those for whom English as an additional language, and those expecting a lower grade are willing to pay more. Purchase likelihoods and essay valuations decline as the probability of cheats being detected, and the penalties if caught, increase.

There's probably nothing too surprising there. However, why is cheating through buying essays a problem? Because it reduces the signalling value of education. As I wrote in this 2014 post:

One of the key characteristics of a degree or diploma is the signal that it provides to prospective employers about the quality of the applicant for positions they have available. Employers don't know up front whether any particular applicant is good (intelligent, hard working, etc.) or not - there is asymmetric information, since each applicant knows their own quality. One way to overcome this problem is for the applicant to credibly reveal their quality to the prospective employer - that is, to provide a signal of their quality. In order for a signal to be effective, it must be costly (otherwise everyone, even those who are lower quality applicants, would provide the signal), and it must be more costly for the lower quality applicants. Qualifications (degrees, diplomas, etc.) provide an effective signal (costly, and more costly for lower quality applicants who may have to sit papers multiple times in order to pass, or work much harder in order to pass).

In the same way that qualifications are a signal, the grade students receive is also a signal of their quality, because it is harder (more costly, in terms of effort) to get an A grade than a C grade. However, if some students are cheating, then high grades are no longer as effective a signal to employers of students' quality. This is because it is no longer more costly for low-quality students to get an A grade, because any student can do so by buying an essay. This reduces the value of education for everyone.

The overall conclusion by Rigby et al. was, unsurprisingly, that if penalties are high enough, students will avoid buying essays. So, universities should be vigilant, and heavily penalise students who are caught cheating. That reduces the expected net benefit of cheating, and reduces the incentive to buy essays. Gary Becker would be proud.

However, an alternative is to make asymmetric information work for you. Most of the online markets where students buy essays simply link up a willing buyer with a willing essay-writer. The sites themselves mostly don't employ people to write essays directly (and those that do are pretty low quality). So, universities could combat this by flooding the online markets with low-quality rubbish essays. How would this work to reduce cheating?

Students can't be sure about the quality of any essay they buy. Essay writers can't easily prove to students that they will write a high-quality essay. So, there is asymmetric information, but is there adverse selection? I argue yes, since sellers of low-quality essays can take advantage of students' inability to distinguish between low- and high-quality essays.

Will the market fail? Students' willingness-to-pay for an essay is affected by the quality of the essay they expect (as per Rigby et al.'s results above). So, if universities flood the market with lots of low-quality rubbish essays, then students will start to expect lower quality essays and adjust their willingness-to-pay downwards, and may drop out of the market entirely (why bother trying to buy an essay if the quality is highly likely to be rubbish). Sites selling essays will make less money and begin to shut down, since they can't easily prove to students that their essays are high-quality. It probably won't make the market fail completely, [*] but it would reduce the problem and be extremely funny (and no doubt distressing for cheating students).

*****

[*] Readers of a certain vintage will remember the music sharing service Napster. In the last months before Napster was shut down, the music companies (I assume - who else would do this?) started flooding the service with fake MP3 files. This didn't work in terms of shutting Napster down, but it was pretty frustrating for users.