Thursday, 30 September 2021

Alarmism about global food prices misses an important part of the story

I was surprised when I read this article in The Conversation this week, by Alastair Smith (University of Warwick):

Global food prices shot up nearly 33% in September 2021 compared with the same period the year before. That’s according to the UN Food and Agriculture Organisation (FAO)‘s monthly Food Price Index, which also found that global prices have risen by more than 3% since July, reaching levels not seen since 2011.

The Food Price Index is designed to capture the combined outcome of changes in a range of food commodities, including vegetable oils, cereals, meat and sugar, and compare them month to month. It converts actual prices to an index, relative to average price levels between 2002 and 2004. This is the standard source for tracking food prices – nominal prices, as they’re known, which means they’re not adjusted for inflation.

While nominal prices tell us the monetary cost of buying food in the market, prices adjusted for inflation (what economists call “real” prices) are much more relevant to food security – how easily people can access appropriate nutrition. The prices of all goods and services tend to rise faster than average incomes (though not always). Inflation means that not only do buyers need to pay more per unit for food (due to its nominal price increase), but they have proportionately less money to spend on it, given the parallel price increases of everything else, except their wages and other incomes.

Back in August, I analysed the FAO’s inflation-adjusted Food Price Index and found that real global food prices were actually higher than in 2011, when food riots contributed to the overthrow of governments in Libya and Egypt.

The idea that "the prices of all goods and services tend to rise faster than average incomes" is flat out wrong. If it were true, then real wages would be declining, and so would living standards. But they're not (see the figures on pages 31-32 of the latest ILO Global Wage report for data covering 2006-2019). It gets worse though. On his blog (which The Conversation article links to), Smith wrote (emphasis in the original):

Sadly, while more precise than previous accounts, this messaging still misses the simple statistical observation that obliterates the relevance of this coverage. Reviewing the “real” price of food over time – expressed as an index relative to a base year, rather than the nominal value in currency, which makes comparison harder due to inflation – the only relevant way to capture today’s status of global food prices is to say that:

‘It is on average harder to buy food today in 2021, than it has been since 2012, and in fact for most of the noughties, the entire decade of the 1990s, and the 1980s; most of the 1970s, and every year of the 1960s! Food is more expensive today than it has been for most of the modern recorded history’.

Not only is it wrong, it's alarmist and wrong. It doesn't even take much economic sense to see why it is wrong. Consumers' decisions about how much food to buy depend on three things: (1) food prices; (2) the consumers' incomes; and (3) the consumers' preferences. If consumers' preferences haven't changed, then the amount of food that consumers buy depends on prices and their income.

If food prices go up, then consumers will buy less food. This is essentially the argument that Smith makes. However, if consumer incomes go up, then consumers will buy more food (because food is what economists call a normal good - that's a good that, by definition, consumers buy more of when their income increases). So, if food prices go up and consumer incomes go up, then consumers may buy less food, or they might buy more food. It crucially depends on how much consumers respond to the change in price, and how much they respond to the change in income. I'll come back to this point using an economic model a little bit later in this post.

But first, on his blog Smith has a figure that demonstrates that the Global Food Index (which tracks the real price of food, and is available here) is about 16.7 percent higher in 2021 than it was in 1961. However, global real gross national income (GNI) per capita (a measure of income per person) more than doubled between 1971 and 2019 (see the World Bank data here). Here's a graph that compares both series (GNI per capita, and Global Food Index) in terms of increases relative to 1971 (when the GNI data series starts). It is clear that incomes have risen far more than food prices over that time. So, it's hard to see how the data supports a claim that "It is on average harder to buy food today in 2021, than it has been since 2012... most of the 1970s, and every year of the 1960s!". Such hyperbole is completely unwarranted.

The comparisons that are made do matter. Food prices in 2021 are 21 percent higher than the average of 2014-2016, but incomes haven't grown nearly as much over that time period (GNI per capita increased by 8.8 percent from 2014-2019, and won't have grown enough in the past two years to match the food price increase). That change over recent years might support Smith's rhetoric (although the alarmist tone is arguable), but his use of the longer time period clearly does not. However, it's not really global food prices that matter for the destabilisation of societies, but local food prices in those societies. Smith could have crafted a much more compelling story by focusing on local changes rather than global.

Anyway, now we come to the economic model I promised earlier (which is the consumer choice model, and useful for ECONS101 students). This model is illustrated in the diagram below, which shows food on the x-axis, and 'all other goods' on the y-axis. The consumer buys a bundle of goods that is made up of food, and all other goods. When the price of food is low (PF0), the black budget constraint applies. The optimising consumer's best affordable choice is the bundle of goods E0, since it is on the highest indifference curve the consumer can reach (I0), while remaining within their feasible set (on the budget constraint). The consumer is buying F0 food. When food prices increase (to PF1), the budget constraint pivots inwards to the blue line and becomes steeper. The consumer can no longer afford the bundle of goods E0 (it is outside of the feasible set), so their new best affordable choice is the bundle of goods E1, since it is on the highest indifference curve the consumer can now reach (I1), while remaining within their new feasible set (i.e. on the new budget constraint). The consumer is worse off, because they are now on a lower indifference curve (which means that they receive less utility, or satisfaction, from their consumption), and they buy less food (F1).

When the price of food increases, there are really two things going on for the consumer. First, there is a substitution effect. Food has become relatively more expensive than all other goods, so the consumer buys less food (and substitutes some of their consumption to all other goods instead, because they have become relatively cheaper). Second, there is an income effect. The consumer's real income (or purchasing power) has decreased - they can afford to buy less in total than they could before. Since food is a normal good, the consumer buys less of it (and since the composite 'all other goods' is also a normal good, they also buy less of all other goods as well - the income effect explains why the increase in the price of food causes the consumer to buy less of all other goods, as well as less food). [*]

However, we are not done. Remember that, in addition to increasing food prices, consumer incomes have increased. If we compare the 1970s with today, the increase in income is far larger than the increase in food prices. This is illustrated in the diagram below (which also shows the situation from the diagram above). The red budget constraint shows the combined effect of the increase in prices to PF1 (the budget constraint becomes steeper) and the increase in consumer income to M2 (the budget constraint moves outwards, parallel to the budget constraint with the new food price). The consumer could have kept buying the bundle of goods E0. They won't though, because they can now reach a higher indifference curve (I2), by buying the bundle of goods E2. The consumer is much better off than before, because they are on a higher indifference curve (which means that they receive more utility, or satisfaction, from their consumption), and notice that the net effect is no change in food purchases at all (the consumer still buys F0 food).

Now consider the change over the more recent period from 2014-2021. This is shown in the diagram below. The food price has increased and consumer income has increased, but the increase in consumer income is now much less (to M3). The consumer cannot continue to buy the bundle of goods E0, because it is outside of the feasible set. Their new best affordable choice is the bundle of goods E3, since it is on the highest indifference curve the consumer can now reach (I3), while remaining within their new feasible set (i.e. on the new budget constraint). Their utility has decreased, as they are now on a lower indifference curve than before, and they buy less food (F3) than before.

Changes in food prices do matter for consumer utility (or satisfaction), and a big decrease in utility will make consumers unhappy. Perhaps a big decrease in utility will make consumers unhappy enough to cause civil unrest. However, food prices are not the only thing that matters for consumer utility. Consumers' incomes matter as well, and if consumer incomes increase faster than (or at least keep pace with) food prices, then it is difficult to see why that would be problematic. Focusing on food prices alone misses an important part of the story.


[*] We could show the income and substitution effects on the diagram, but it isn't necessary for us to do so in order to explain them.

Wednesday, 29 September 2021

Prohibition, land values, and productivity

When we think of prohibition in the U.S., many of us will think about the negative impacts that the black market trade in alcohol has. We think about Al Capone and tommy guns. However, national prohibition (which came into force on 17 January 1920, one year after the 36th state ratified the 18th Amendment to the Constitution) was actually the endpoint of decades of local prohibitions, which were implemented at the county (or sometime the sub-county) level. By 1920, the U.S. was a patchwork of wet counties (where alcohol sales were allowed) and dry counties (where it was not).

Although we mostly think about the negative effects, could local prohibitions have positive effects as well? A recent working paper (forthcoming in the Journal of Economic History) by Greg Howard (University of Illinois) and Arianna Ornaghi (University of Warwick) addresses that question. Specifically, they use Population Census and Agricultural Census data to look at differences in land values (proxied by farm values and population change) and a number of other variables, between counties that were early adopters of prohibition (between 1900 and 1909), and those that were later adopters (1910-1919). The outcomes they look at are measured over the period 1890 to 1910, so the 'late adopters' are counties that hadn't (yet) implemented dry laws, but were similar to the counties that had, in the sense that they were the next counties that would go dry (and would all do so before national prohibition was introduced).

Howard and Ornaghi find that:

...local Prohibition had significant economic effects on rural counties. First, Prohibition increased population and land prices, consistent with it being a policy that people find desirable. Second, we show that counties that enacted Prohibition saw increases in labor productivity and capital investment after they became dry, consistent with agglomeration that comes through a land price channel. We also see an increase in banks in the areas, suggestive of more lending... Third, we show counter-intuitive sorting patterns: counties with local Prohibition attracted relatively more immigrants and African-Americans. Given that these groups were generally less in favor of Prohibition, these sorting patterns seem unlikely to have been driven by preferences for the policy, but could have been the product of growing labor market opportunities.

A model of Tiebout competition suggests that local areas that adopt attractive policies will, by definition, attract population. That will increase land values. Howard and Ornaghi's evidence suggests that Prohibition was a 'positive amenity' that attracted population. The increase in land prices gave land owners more collateral, which they could use to borrow and invest in their farms (remember that this time period was when mechanisation of farms was increasing). The increase in bank lending they identify is consistent with this.

So, local prohibition had positive amenity effects. However, we need to be cautious about extrapolating too much from the U.S. experience a century or more ago. Although alcohol has large and negative effects now (see my post from earlier this week), those effects were likely much larger in the early 20th Century than they would today. So, local prohibition likely had a much larger positive effect than a similar policy would have today, and to some extent we can see that with the benign effect of local alcohol policies in New Zealand, as I discussed earlier this week. Of course, that isn't the point that Howard and Ornaghi are making with their paper - they are more focused on the effects of local amenities and on Tiebout competition, and Prohibition simply provides the context for their study.

[HT: Marginal Revolution, last year]

Read more:

Monday, 27 September 2021

Local alcohol policies and crime in New Zealand

When the Sale and Supply of Alcohol Act 2012 came into effect, it gave local councils the ability to enact enforceable local alcohol policies, where they could impose more stringent conditions on holders of alcohol licences. Councils could require shorter opening hours, or restrict the locations of new alcohol outlets. Nine years on, not all councils have local alcohol policies in place. For those that don't, the default provisions from the Sale and Supply of Alcohol Act apply.

The object of the Act is that "the sale, supply, and consumption of alcohol should be undertaken safely and responsibly; and the harm caused by the excessive or inappropriate consumption of alcohol should be minimised". A reasonable question, then, is do local alcohol policies reduce harm? That is essentially the research question that this 2020 working paper by Lauren Tyler-Harwood and Andrea Menclova (both University of Canterbury) addresses. Specifically, they look at monthly crime rates from July 2014 to January 2019 (the first local alcohol policy didn't come into force until August 2014), and compare territorial authorities with different levels of stringency of local alcohol policies (with the least stringent being those that had no local alcohol policy). Over the period they analyse, 32 of the 66 mainland territorial authorities implemented a local alcohol policy, and there was lots of variation in the specific policy elements that were included. The policy elements Tyler-Harwood and Menclova investigated were: (1) a more stringent on-licence closing time (the default is 4am); (2) a one-way door policy; (3) club licence closing times that are earlier than on-licence closing times; (4) a restriction on issuing new licences; and (5) the difference between on-licence and off-licence closing times.

In terms of the overall effect of a local alcohol policy, they find: statistically significant relationship between adopting an LAP and crime.

Looking at the different policy elements, they find:

...little evidence that crime rates fall more in TAs with more stringent LAPs. The only LAP dimension that is consistently negative and statistically significant at the 1% level is the 11pm maximum on-licence closing time. On average, and holding all else constant, adopting a maximum on-licence closing time that is five hours earlier than the national default is associated with a 6% decrease in monthly crimes. However, given that only one TA, Waimakariri, has adopted this closing time (and only for weekdays), and that there is not a consistent pattern of increasingly strict on-licence closing hours being increasingly negatively associated with crime, we interpret this result with great caution.

Then, looking at different types of crime, they find that:

...the introduction of an LAP is associated with a 5% decrease in assaults... However, this result is only significant at the 10% level. We fail to find significant relationships between LAP policies and other types of crime.

Overall, there is very weak evidence that local alcohol policies have had any effect on crime at all. Now, of course, crime is not the only alcohol-related harm that a local alcohol policy would be implemented in order to address. However, crime is among the most acute harms caused by alcohol consumption (in terms of causality here, see yesterday's post).

We could be tempted to interpret these results as disfavouring local alcohol policies. However, there are good reasons not to do so. First, these results are not causal. Local alcohol policies are not implemented randomly, and Tyler-Harwood and Menclova don't apply statistical methods that attempt to extract causal estimates (and, to be fair, they don't claim to be intending to do so). Second, there isn't much of a baseline period included in the data that Tyler-Harwood and Menclova use. Without a longer baseline, it is difficult to establish a suitable counter-factual for what would have happened in the absence of a local alcohol policy.

However, more importantly, it pays to consider what the local alcohol policies that have actually been implemented have entailed. Local alcohol policies with more stringent policy elements generated a lot of push-back from the alcohol industry. Many local alcohol policies went through a long appeals process, and this was costly for local councils. Many councils (including Hamilton City Council) opted to withdraw their provisional local alcohol policies in the face of industry opposition, and the prospect of a lengthy and expensive court battle. Knowing how the industry would react made it more difficult for councils to implement policies with more stringent restrictions on the sale of alcohol.

So, it is not surprising that there are no statistically significant effects of local alcohol policies on alcohol-related harm. The policies that have been implemented have generally been too benign to have much of an effect (on a related note, read this editorial (possibly gated) in the journal Drug and Alcohol Review by Nicki Jackson and Kypros Kypri on the Auckland local alcohol policy).

If local alcohol policies are to have a significant effect on alcohol-related harm, then they will need to actually be stringent enough to shift alcohol consumption behaviour. We don't have that at the moment. However, most of the initial round of local alcohol policies will be coming up for review in the next couple of years. It will be interesting to see what evolves, and whether any councils will attempt more stringent controls.

Sunday, 26 September 2021

Alcohol, crime, and long baseball games

There are a number of theories that link alcohol consumption with crime. One that I have relied on in much of my own research is called routine activity theory. This theory suggests that crime occurs when you have motivated offenders and attractive targets (or potential victims, in the case of inter-personal crime) in the same location, in the absence of capable guardians. Under this theory alcohol is a 'chemical facilitator' of crime, because it reduces the inhibitions of the offenders, and impairs the victims. Routine activity theory comes from human ecology, and there are lots of alternative theories from criminology that can be used to support the link from alcohol consumption to crime. However, they are just theories. The evidence to support them is almost entirely based on observational studies. There is precious little that demonstrates causation, rather than correlation, between alcohol consumption and crime.

A new article by Jonathan Klick and John MacDonald (both University of Pennsylvania), published in the Journal of Quantitative Criminology (may be ungated, but just in case there is an earlier version here), does provide some causal evidence, using a cool identification strategy.

Klick and MacDonald look at crime occurring around Citizens Bank Park (CBP), the home stadium of the Philadelphia Phillies baseball team. At baseball games at CBP, alcohol sales are ceased after the seventh inning. From that point on, the baseball fans will become increasingly sober. The longer the baseball game goes on, the more sober they will become. Given that baseball is not a time-limited game, there is a lot of variation in the time between the seventh inning and when the game finishes and the fans spill out of the stadium (where, as routine activity theory suggests, problems may occur). Klick and MacDonald make use of that variation and look at how crime varies depending on the length of the baseball game. As they explain:

...we rely on a novel natural experiment to provide causal estimates of the impact of alcohol consumption during Major League Baseball (MLB) games on crime near a stadium... The game duration from the end of the seventh inning to the end of the game can be short or long... This aspect of MLB games allows us to examine a wide range of time spans during which spectators are limited in their ability to drink alcohol. Thus, we can compare game days with non-game days when the game is at home or away, and when the time from the end of the seventh inning extends allowing fans who are in attendance more time to sober up.

Using data from 2006 to 2015, they find that:

...home games that are relatively lengthy after the seventh inning and games with extra innings generate lower crime around CBP, as compared to other areas around the city. For the average game, it appears that the alcohol sales restriction reduces assaults by 40 to 70 percent. These effects are concentrated within the first hour after the game, with little additional crime reduction occurring after that, and in a relatively small area around CBP. We do not observe similar effects around a selection of popular sports bars in other areas of the city where no seventh inning restriction on alcohol sales applies.

The comparison with the areas around sports bars is an important one. Of course, sports bars do not have to cease alcohol sales at the seventh inning, so patrons leaving the bars will be more intoxicated than those leaving CBP. Also, there is a further natural experiment that Klick and MacDonald can exploit:

In March 2012, the Xfinity Live!... complex opened in the stadium parking lot. This entertainment venue contains several bars and restaurants that sell alcohol until 2 a.m. each evening, effectively undoing any potential effect of the alcohol sale stoppage in the stadium at the end of the seventh inning.

As expected, they find that:

These effects largely vanish after the Xfinity Live! complex opened and allowed fans to continue to drink alcohol after the seventh inning in the stadium parking lot, further suggesting the link between the stadium alcohol restrictions and crime is causal.

Klick and MacDonald test their analysis based on various geographies representing the size of the area of effect of stadium drinking, and find similar effects. While it might seem obvious that there is a causal link between alcohol consumption and crime, the actual empirical evidence to support that link is limited. This study goes some way to filling that research gap.

[HT: Marginal Revolution]

Read more:

Saturday, 25 September 2021

Coronavirus, excess demand, and the NHS waiting list

iNews reported last month:

NHS England’s waiting list could rise to 14 million by autumn next year and continue increasing due to the backlog caused by the Covid pandemic, a report has warned.

Millions of patients were not able to receive care during the crisis, and if they returned to the NHS, the number of those joining the waiting list could overtake those being treated, the Institute for Fiscal Studies (IFS) said.

How did the NHS end up in this situation, with tens of millions of patients waiting for care? Like New Zealand, the NHS is a public health system. Under a public health system, health care is funded by the government (out of general taxation), and provided to patients at no cost (or costs might be shared with patients paying a low co-payment for some types of care).

However, a public health system doesn't overcome the fundamental scarcity of health care resources. In a private health system, the price of health care determines who does, and who does not, receive health care (those who are willing and able to pay the market price, often with the assistance of private health insurance, will receive the scarce health care). In a public health system, the price of health care is low (or zero), and so price does not ration the scarce health care resources. Instead, health care must be rationed some other way. In England (and New Zealand), the health care is rationed using a waiting list. If you want health care, you have to wait for your turn.

To illustrate this, consider the market for health care, as shown in the diagram below. If this was a private health system, the equilibrium price (P0) would prevail, and Q0 health care services would be delivered. However, in a public health system, the market price is zero. At the zero price, the quantity of health care services demanded is QD0, while the quantity of health care services supplied is QS0. There is not enough health care services to satisfy the demand, and this excess demand (or shortage) of health care services (which is equal to [QD0-QS0] is managed by a waiting list.

Now, consider how the coronavirus pandemic changed things in England. This is illustrated in the diagram below. Due to people getting COVID-19, the demand for health care increased from D0 to D1. At the same time, the supply of health care decreased from S0 to S1 (that's what happens when health care providers get sick as well!). With the price of health care stuck at zero, the quantity of health care demanded increased to QD1. The supply of health care decreased to QS1. So, the excess demand for health care increased from [QD0-QS0] to [QD1-QS1]. The waiting list increased.

How can the NHS avoid this situation? The iNews article says:

Estimating the number of returning patients and the capacity of the NHS, the IFS said the most optimistic scenario would see 80 per cent of ‘missing’ patients return to the NHS.

This would increase waiting times to nine million next year, which would not drop back to pre-pandemic levels until 2025.

To achieve this, the NHS will have to increase its capacity by five per cent over the next two years, compared to 2019, and then by 10 per cent from 2023.

Increasing the supply of health care services (by increasing the capacity of the health system) would shift the supply of health care back to the right. However, notice that the waiting list would not be eliminated, only reduced. Health care resources remain scarce. Alternatively (and not noted in the article), vaccinations reducing the demand for health care would also reduce the waiting list. The key point here is that, if changes in the demand for (and supply of) health care can't affect the price (as is the case in a public health system), then they will be reflected in the length of the waiting list.

[HT: Marginal Revolution]

Thursday, 23 September 2021

The shearer shortage and the price of winter woollies

The New Zealand Herald reported this morning:

A shortage of shearers has cost farmers this coming summer, with kiwi and Aussie shearers stuck on the other side of the Tasman due to closed borders.

It's not just shearers but also shed hands and wool handlers that could be in short supply.

A shortage is a situation where the quantity demanded exceeds the quantity supplied. That means that the current market price is below the equilibrium price. That is illustrated in the diagram below. The market price of shearing services is P0, which is below the equilibrium price of P1. [*] At the market price, the quantity of shearing services demanded is QD, but the amount of shearing services available (the number of shearers wanting to work) is only QS. The difference between QS and QD is the shortage of shearers.

What happens next? If you are a sheep farmer, and you want your sheep shorn, there is a good chance that you aren't able to find a shearer because of the shortage. One way for you to solve the problem is to contact a shearer, and offer them more money to shear for you, so that you won't miss out. Other farmers are going to be doing the same though, so effectively, the sheep farmers bid up the price of shearing services. This continues all the way to P1, where the quantity of shearing services supplied is equal to the quantity of shearing services demand (and both are equal to Q1). Interestingly, this is exactly what the New Zealand Herald article anticipates:

With a looming labour shortage farmers could see the price per sheep go up compared with last season.

"Obviously people have got to pay what they have got to pay to get people and you are all fighting over a small labour market"

The increase in the price of shearing services will have a flow-on effect onto the market for wool. This is shown in the diagram below. The supply curve shows the costs of production of the sheep farmers. An increase in the price of shearing services increases the costs of production of wool, and shifts the supply curve up (and to the left) - what we refer to as a decrease in supply. The equilibrium price of wool will then increase from P0 to P1 (and the quantity of wool traded will decrease from Q0 to Q1).

This then flows onto the costs of production of items that are made of wool. Higher wool prices increase the costs of production, decreasing supply and increasing the price (the diagram is the same as the one for the wool market, above). So, the shearer shortage is going to cause your winter woollies to increase in price. Thankfully we're heading out of winter!


[*] Arguably this is a labour market, and so we should refer to the price as a wage. However, since most shearing is undertaken by contract firms, I've opted to show this as a services market instead. The implications are the same regardless of whether this is a market for services or a labour market.

Wednesday, 22 September 2021

Demand and the price of NFL team branded face masks

The demand for face masks in New Zealand remains high, with face masks required to be worn in most indoor spaces in Level 2 and Level 3 lockdown. However, the high demand New Zealand is experiencing isn't replicated everywhere around the world, and we can actually benefit from it.

Yesterday, while online shopping for an NFL jersey (go Panthers!), I saw this amazing deal:

Yes, that is a pack of three branded face masks for US$1.99, down from the 'regular' price of $24.99. Naturally, I bought some. However, it is worth thinking about why the price dropped so dramatically.

In the simplest terms, it comes down to supply and demand (or rather, just the change in demand). When masks were required in some states in the US, then it made sense for some NFL fans to buy masks branded with their team logo. However, now that masks are no longer required, demand has fallen dramatically. This is illustrated in the diagram below. When demand was high (D0), the equilibrium price of masks was high (P0 = $24.99). But now that demand is lower (D1), the equilibrium price of masks has fallen (to P1 = $1.99).

The NFL shop isn't lowering prices out of a sense of goodwill towards consumers. And no one is forcing them to lower prices. Clearly, the NFL shop found themselves overstocked and is looking to get rid of their excess stock. Lowering prices means that they sell more than if they had kept the price high - if they had kept the price at P0 after demand had fallen to D1, the quantity demanded would have fallen all the way to Qd. The result of NFL shop trying to move face masks that there is little demand for in the US is a ridiculously low price for everyone.

The lower price is an especially great deal for those of us in countries where masks are still required. We get to pay the low U.S. price, even though our demand is still high. That's what I'd call maximum transaction utility.

Tuesday, 21 September 2021

Scott Cunningham on the mental health of PhD students in economics

Mental health of students is a recurrent theme of late (e.g. see here). It was a key theme of research that Barbara Fogerty-Perry (Otago Polytechnic), Gemma Piercy Cameron (University of Waikato), and I did, looking at the impacts of the pandemic on tertiary students in New Zealand. Our sample was mostly made up of undergraduate students. However, students undertaking independent study, such as PhD students, may be more of a worry. Doctoral study can be a lonely and isolating experience.

So, I was interested to read this recent substack post by Scott Cunningham on depression, anxiety, and suicidality among economics PhD students. The post is based on two studies: (1) this meta-analysis of mental health among PhD students (in all fields, and worldwide), by Emily Satinsky (Massachusetts General Hospital) and co-authors (including Scott Cunningham), published in the journal Scientific Reports (open access); and (2) this forthcoming paper in the Journal of Economic Literature (will be ungated when published, but for now there is an ungated earlier version here), by Valentin Bolotnyy (Stanford University), Matthew Basilico, and Paul Barreira (both Harvard University), on mental health among economics PhD students.

Cunningham's post is difficult to excerpt, since there is a lot of detail in it. However, here is one part:

Students are our in our care and therefore we bear a certain amount of responsibility to these mental health struggles. It is likely that these mental health struggles are endogenous, not only to the students own choices, but to our personal and collective choices as well. They are young, they are working hard, and their struggles are often unknown to the very people, like advisers and friends, who might be in a position to help. The first step is to learn more about the pervasiveness of these problems. I encourage you read these two papers with curiosity and an open mind.

I encourage you to read the whole of Cunningham's substack post, and follow up with the research articles if you are interested. Since Cunningham does such a good job of summarising the two articles, I'm not going to repeat the effort here, other than to note these key statistics from the Bolotnyy et al. paper:

We find that 17.7% of students are experiencing moderate or severe symptoms of depression, 17.6% are experiencing such symptoms of anxiety, and 24.8% are afflicted with one or the other. These rates are 2 to 3 times the national prevalence, but are similar, if not lower than, estimates produced by other studies of graduate student mental health...

The takeaway from that is that the situation is bad in economics, but potentially no worse in economics than in other fields of PhD study. Bolotnyy et al.'s study is based on eight high-ranked PhD programmes (Columbia University, Harvard University, University of Michigan, Massachusetts Institute of Technology (MIT), Princeton University, UC Berkeley, UC San Diego, and Yale University). It would be interesting to see if the situation was as bad at lower-ranked, less selective institutions. It could be better, or it could be worse - it depends on whether positive selection of students on the basis of academic performance negatively selects them on measures of mental health. It also depends on the extent to which the pressures of PhD study cause worse mental health (neither of the cited studies can help with that, since they are cross-sectional and so only show correlations). It also depends on how resilient PhD students are, and how that correlates with academic performance (and neither study addresses resilience).

Clearly, there is more to learn on this topic. It was instructive to learn that, from the Satinsky et al. meta-analysis, there has not been a study of PhD student mental health in New Zealand (at least, not one that measured depression or anxiety), and only a couple in Australia. Given that the New Zealand model of PhD study, with little coursework and students working on a single, large research project (or several, smaller, inter-connected projects), differs substantially from the US model, it may be that there are differences in the consequences for PhD student mental health. Again, mental health among New Zealand PhD students could be better, or it could be worse. Perhaps the coursework component creates a cohort of colleagues that PhD students connect with, who can help them through their studies, improving mental health outcomes. Or, as I said at the outset of this post, perhaps the solitary research project approach is intensely isolating and worsens mental health. Either way, the current pandemic, where international PhD students are commencing their New Zealand PhD while still overseas, distant from supervisors, peers, and more senior PhD students, is going to be incredibly challenging, both in terms of work and mental health. I really worry for the two PhD students who I have who are currently in this position (and hope that the borders open soon to let them get over here into what is a, if not fully supportive, at least less isolating, environment).

Anyway, coming back to the research papers that Cunningham discusses, Bolotnyy et al. make six recommendations that are worth considering:

First, we recommend that department leaders raise awareness of mental health issues among graduate students, raise awareness of available mental health resources, and encourage students to take their mental health and the health of their peers seriously...

Second, department leaders could use their platform to encourage students to invest in building friendships with their peers and to actively avoid prolonged isolation...

Third, improving student-faculty advising relationships can help students identify promising directions for research and bounce back better from setbacks... Some departments have started connecting students early on with faculty who volunteer to advise students in the pre-research years. Such advising relationships, established outside of the dissertation committee structure, may provide students with faculty support that does not come bundled with consequential evaluation...

Fourth, relatedly, we recommend instituting policies that help advisers ensure that students are not falling through the cracks and are progressing with their projects. In programs where the advising structures are more diffuse, field-specific meetings among faculty to discuss student progress could be a good way to do this...

Fifth, with so few students finding meaning in their work, we think it would be useful to actively encourage students to pursue research questions they find meaningful and socially valuable...

Sixth, departments could partner with campus mental health services to experiment with different approaches to mental health treatment. Some departments have experimented with peer support groups and "Let's Talk" programs that make campus mental health professionals available for drop-in hours close to the department...

I absolutely agree on raising awareness, encouraging students to build friendships and peer networks and to engage with staff members other than their supervision team on a regular basis, and building better links between academic programmes and mental health services (this applies at all levels of study). Regular meetings of PhD supervisors might be helpful in some circumstances.

Two of the recommendations (#3; and #5) strike me as a little problematic though. In relation to having non-supervisors "provide students with faculty support that does not come bundled with consequential evaluation", we should recognise that this type of emotional labour is more likely to fall on the shoulders of female faculty members, potentially exacerbating gender differences in workload and research performance. There needs to be mechanisms in place to ensure that doesn't happen, before this recommendation will be feasible to implement. Also, there needs to be clear expectations about professional boundaries, to avoid other problems arising (the Bolotnyy et al. paper has a whole section on sexual harassment, for example).

In relation to pursuing research questions that are meaningful and socially valuable, those research questions are unfortunately often not valued as highly within the field as questions that are less meaningful but more publishable. This is a problem that will be most difficult to address, as PhD students who have aspirations of tenure and/or a position at a high-ranking institution will continue to prefer to study research questions that are publishable in the top journals. It would require the top journals to be more open to other research questions, or are-organisation of the journal publishing model (see Maximilian Kasy's recommendation in this post, for example).

PhD student mental health is clearly a major problem. These studies (and Cunningham's post) have highlighted just how bad the problem is. However, we still need to know more, and to develop (and test) potential solutions.

[HT for the Cunningham post: David McKenzie at Development Impact]

Monday, 20 September 2021

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.

Friday, 17 September 2021

Book review: Globalization and Its Discontents

Joseph Stiglitz won the Nobel Prize in 2001, for his work on information asymmetry and screening. However, before winning that prize he was on Bill Clinton's Council of Economic Advisors from 1993 to 1997, the last two years as Chairman. After that, he went to the World Bank to be their senior vice president for development policy. That was just before the Asian Financial Crisis hit. Stiglitz remained at the World Bank until 2000, when he was ousted by US Treasury Secretary Lawrence Summers. Then, he wrote a book about his experiences at the World Bank, Globalization and Its Discontents.

Despite the title, the book isn't a take-down of globalisation in its entirety. As Stiglitz explains:

Globalization itself is neither good nor bad. It has the power to do enormous good, and for the countries of East Asia, who have embraced globalization under their own terms, at their own pace, it has been an enormous benefit, in spite of the setback of the 1997 crisis. But in much of the world it has not brought comparable benefits.

However, Stiglitz is clear about who has benefited the most from globalization, noting that:

...the West has driven the globalization agenda, ensuring that it garners a disproportionate share of the benefits, at the expense of the developing world.

So while globalisation overall doesn't get a roasting from Stiglitz, the same cannot be said of western countries in general, and the International Monetary Fund (IMF) and the US Treasury in particular (so it's no wonder that Summers was displeased). The IMF has rightly come in for a lot of criticism from many quarters over the years for its handling of financial crises, developing country debt, and its single-minded market fundamentalism. Stiglitz doubles down on all of that, and more. For example, here are some key quotes:

But the IMF did not want to take on the mere role of an adviser, competing with others who might be offering their ideas. It wanted a more central role in shaping policy. And it could do this because its position was based on an ideology - market fundamentalism - that required little, if any, consideration of a country's particular circumstances and immediate problems...

The Fund's economists have never laid claim to being great theorists; its claim to expertise lay in its global experience and its mastery of the data. Yet strikingly, not even the data supported the Fund's conclusions...

...the perception throughout the developing world, one I share, is that the IMF itself had become a part of the countries' problem rather than part of the solution...

...intellectual consistency has never been the hallmark of the IMF...

Ouch! Stiglitz draws on examples from several financial crises including the Asian Financial Crisis, Russia, and debt crises in sub-Saharan Africa. The book was an interesting read about a period of global financial turmoil that has since been significantly overshadowed by the Global Financial Crisis of 2008. You would hope that the IMF has reformed its ways in the twenty years since the events in the book. However, they are still being criticised for similar problems.

While Stiglitz clearly has no time for ideology (as I've said before, ideology is the result of lots of people suffering from loss aversion and the endowment effect), he does identify a more important driver of the IMF's focus on markets: the revolving door between the IMF and international banks and other financial institutions. In particular, Stiglitz notes that it is in IMF officials' best interests to ensure policy and bailouts benefit western financial institutions:

Simplistic free market ideology provided the curtain behind which the real business of the "new" mandate could be transacted. The change in mandate and objectives, while it may have been quiet, was hardly subtle: from serving global economic interests to serving the interests of global finance.

By cosying up to global finance, IMF officials could ensure themselves a future high-paying job in the institutions that benefited most from IMF bailouts and the conditionality placed on countries receiving IMF loans. This is a problem that most countries face with politicians as well, unless there are rules to prevent this 'revolving door'.

One other interesting thing struck me in the book, which was the lack of transparency of negotiations between the IMF and countries seeking assistance. This struck me as eerily similar to the negotiation of free trade deals, where the electorate (and even elected representatives) are presented with a finalised deal as a fait accompli, with limited or no opportunity for genuine debate or deliberation. On negotiations with Russia, Stiglitz writes:

These are complicated matters, and in democracies, they need to be debated and discussed. Russia was trying to do that, trying to open up the discussion to different voices. It was Washington - or more accurately, the IMF and the U.S. Treasury - that were afraid of democracy, that wanted to suppress debate. I could not but note, and feel sad about, the irony.

Things have gotten pretty bad when someone makes a comparison in terms of transparency, between your organisation and a country like Russia, and Russia comes off looking better.

This book has aged well, and unfortunately too many of the critiques remain valid. It also provides some great insight into the problems that developing countries face at a macroeconomic level, when dealing with donor nations that have all of the funds and all of the power. That in itself still makes it a relevant read nearly twenty years after it was first published (and interesting, I see that there is a new revised and expanded edition available that revisits the issues, published in 2017).

Thursday, 16 September 2021

Should New Zealand have a coronavirus vaccination lottery incentive?

Vaccination incentives have been in the news again this week, as steel building product manufacturer Steel and Tube began paying its workers $150 to get vaccinated, and Auckland Airports Park & Ride vaccination centre began daily prizes worth $70,000 in total. And in terms of extreme incentives, you have both Air New Zealand and Ryman Healthcare proposing policies to make vaccinations mandatory for all of their employees.

Vaccination incentives have been undergoing debate in the 'pages' of The Conversation over the last few months (see here and here and here). It's also a topic that I have commented on before (see here and here). Both of my posts referenced the Vax-A-Million lottery incentive in Ohio. So, I was interested to read this recent working paper by Andrew Barber and Jeremy West (both University of California, Santa Cruz), that evaluates the impact of the Ohio lottery on coronavirus vaccinations, cases, and hospitalisations.

The Ohio lottery began on 12 May 2021 and ran until 20 June 2021, and Barber and West use daily data covering the period from 19 February to 18 July. They compare Ohio with a synthetic control, which is derived from shares of other states (mostly Wisconsin, Kansas, Michigan, Idaho, and North Dakota) that had a similar trajectory prior to the lottery for both cases and relevant control variables. They find that there is: increase in COVID-19 vaccinations in Ohio that begins almost immediately after the Vax-A-Million announcement and persists past the final prize drawing. Relative to the synthetic control, the program causes a 0.7 percentage points (1.5 percent) increase in the share of state population receiving at least a first dose of a COVID-19 vaccine by the program’s end date, with most of this effect occurring within two weeks of the announcement. In levels, this amounts to about 82,000 people who were persuaded to vaccinate by the CCL incentive, implying an average program cost of 68 dollars per “complier.” For context, this cost-per-complier is less than the 80 dollars in direct costs that the federal government pays a healthcare provider to fully vaccinate one person...

The results are even more compelling when you see them as a picture. Here's their Figure 2 (panel (a)), which shows the difference between the national vaccination rate, and the rates in Ohio and 'synthetic Ohio':

That doesn't look like a big difference, but the trajectory is clearly changed for Ohio (relative to synthetic Ohio) after the vaccination lottery starts. It is easier to see in panel (b) of the same figure, which shows the difference:

The difference in vaccination rates jumps up soon after the lottery starts, and remains higher thereafter. There are similar results for COVID cases, but with a lag (which is to be expected, because the vaccination takes some time to reach full effectiveness). From their Figure 3, panel (b):

And similar again for hospitalisations, with a further lag (because it takes time for cases to get hospitalised). From their Figure 4, panel (b):

You may have read about other research on the Ohio lottery, questioning its effectiveness. Barber and West address that in their paper:

Walkey et al. (2021) conduct an interrupted time series study of Ohio versus the United States during the few weeks surrounding the lottery announcement, concluding that Ohio’s program does not increase vaccination rates. However, Ohio’s vaccination rates track poorly with national rates during the pre-treatment period—a factor motivating our synthetic control identification strategy. Lang et al. (2021) use the classic SCM to study how Ohio’s program affects the share of fully vaccinated residents, finding no effect. This null effect could be because lottery eligibility only required a single dose rather than full vaccination. In addition, their study stops at the final lottery drawing, weeks before many lottery-eligible participants could have obtained a second dose of a vaccine series, which require 21 or 28 day gaps between doses... Finally, Brehm et al. (2021) use county-level data from Ohio, Indiana, Michigan, and Pennsylvania to conduct pooled SCM and state-border difference-in-differences estimations of how Ohio’s program affects the number of first dose vaccinations during the treatment period. The study finds an effect on vaccinations that is very similar in magnitude to that we show here.

What we can take away from this is that the Ohio vaccination lottery appears to have increased the vaccination rate, decreased COVID cases, and decreased hospitalisations. And all at a cost of US$68 per additional vaccination. To get to 90% of people with at least one dose, we need to vaccinate another 800,000 people. At US$68 per vaccination, that's about NZ$76 million. Seems like a steal to me, especially when a Level 3 lockdown in Auckland alone has been estimated to cost $45 million per day. Even if you paid out US$68 on average to everyone, including those that have already been vaccinated, it would still be a good deal.

[HT for the Barber and West study: Marginal Revolution]

Read more:

Wednesday, 15 September 2021

Uncovering labour market discrimination against foreign-born and native-born minority workers

There is an interesting strand of research in economics (especially labour economics) that seeks to uncover evidence of discrimination using what are referred to as correspondence tests. Essentially, the researchers send out pairs of job applications to a bunch of firms, where each pair of applications differs only in some characteristic that the researchers want to test for discrimination against. So, for example, in the original research using this method (ungated version here) by Marianne Bertrand and Sendhil Mullainathan, the CVs they sent out had either an African American name or a non-African-American name. They then compared the number of each type that were invited to interviews.

The literature that applies this approach is large and growing. However, this short 2017 article by Nick Drydakis (Anglia Ruskin University), published in the journal Economics Letters (ungated earlier version here) caught my eye. In the paper, Drydakis compares invitations to interviews at 344 Greek firms between CVs that were designed to be interpreted as being for Greek-born applicants with Greek names (natives), foreign-born applicants with foreign names (non-natives, where the names were obviously Albanian, Ukrainian, or Georgian), and Greek-born applicants with foreign names (natives with an ethnic minority background). He finds that:

...natives with an ethnic-minority background have a 17.5 percentage points lower chance of receiving an invitation for interview than natives. Also, it is observed that non-natives have a 20.1 percentage points lower chance of receiving an invitation for interview than natives. Both estimates are statistically significant at the 1% level. However, the two estimates are not statistically significantly different...

In other words, Greek-born applicants with non-native names and foreign-born applicants are essentially discriminated against to the same extent. That suggests that discrimination in this context is to some extent taste-based discrimination (where employers have preferences not to employ the minority group), rather than statistical discrimination (where belonging to a particular group is statistically associated with lower productivity, which might arise for instance when that group has lower education). We can infer that statistical discrimination is less likely, because the CVs of the Greek-born applicants with non-native names showed the same educational and labour market backgrounds as the CVs of the native applicants.

Also interesting is that the discrimination extends to wages as well:

...natives with an ethnic-minority background are invited for interviews for vacancies that offer 5.5 percentage points lower wages compared to natives. Moreover, the estimates suggest that non-natives are invited for interviews for vacancies that offer 6.4 percentage points lower wages compared to natives. Both estimates are statistically significant at the 1% level. However, the two estimates are not statistically significantly different...

So again, there is no difference between the Greek-born applicants with non-native names and foreign-born applicants. This should be especially disappointing for children of immigrants. I wonder if these results hold in other countries besides Greece? I also wonder to what extent these effects might be moderated by contact with minority groups, as per the contact hypothesis I discussed yesterday. Those suggest some potential avenues for future research.

Read more:

Tuesday, 14 September 2021

The contact hypothesis, and African American GIs in Britain

The contact hypothesis posits that contact between members of majority and minority groups (under certain conditions) can reduce prejudice towards the minority group. In a new test of this hypothesis, this article by David Schindler (Tilburg University) and Mark Westcott (Munich Graduate School of Economics), published in the journal Review of Economic Studies (ungated earlier version here), uses data on African American GIs in Britain. They introduce their paper as:

In this paper, we show that the temporary presence of African American G.I.s... in the UK during World War II persistently reduced anti-minority prejudice amongst the British population. As the base of the US military’s European operation, the UK played host to over one and a half million US troops during World War II. Around 150,000 of these troops were black, serving in segregated units with non-combat support duties such as transport and supply... Many Britons thereby saw and interacted with non-whites for the very first time. Despite pervasive racist attitudes before the war, we show evidence from surveys that these interactions were positive experiences for both the local population and for black G.I.s.

More specifically, they derive a measure of which local areas in England and Wales had military bases where African American support units were stationed. They focus on support units because:

As in previous wars, black soldiers served in racially segregated units, normally under command of white officers. With few exceptions, black troops were limited to non-combat “labour” or “service” roles, most often supply and quartermaster services, transport, food preparation, and sanitation...

Schindler and Westcott then compare areas in England and Wales that were more, or less, exposed to African American troops (controlling for the presence of other support troops) during World War II. They make this comparison in terms of a number of measures that are plausibly associated with racial bias in more recent times, including: (1) membership of the far-right BNP political party (using a membership list published online in 2008); (2) local election results (in terms of the share of votes for the BNP in elections from 2006-2012; or the share of votes for the Conservative Party from 1973-2012); (3) online data from Implicit Attitudes Tests taken by UK residents between 2004 and 2013; and (4) online survey data (from the same source as the IAT data) on 'warmth of feelings' towards African Americans. They find that:

...individuals in areas of the UK where more black troops were posted are more tolerant towards minorities 60 years after the last troops left. First, we show that such areas contain fewer members of the British National Party (BNP), a far-right political party with racist policy positions. Next, we demonstrate that voters in affected areas were less likely to vote Conservative in local elections during times when far-right parties were not widely fielding candidates (until the early 2000s). This effect disappears as the BNP emerged as the strongest far-right party and BNP candidates subsequently received fewer votes in locations where black G.I.s were posted. Finally, we show that there is less implicit anti-black bias in these areas, as measured by Implicit Association Test (IAT) scores, and that those living in locations where black G.I.s were posted report warmer feelings towards black people.

The size of the effects are small. Schindler and Westcott provide some additional analyses where they mathematically manipulate their results to account for the number of generations that have passed between World War II and the present day, and the proportion of the population that might have been exposed to African American GIs. Those adjustments rely on some heroic assumptions, and I don't find them compelling. However, in terms of the contact hypothesis, their analysis separated into rural and urban areas is potentially more important, where they find that:

...the effect of black troops on BNP membership is about twice as large in rural areas as in urban areas. In fact, the effect in urban areas is not statistically significant at any conventional level, despite a larger sample size.

That's consistent with the contact hypothesis because urban areas in England and Wales already had minority populations prior to World War II, so exposure to African American GIs is likely to have a much smaller (in this case, effectively zero) impact. Schindler and Westcott only provide this disaggregated analysis for the BNP membership data - in the other analyses, they simply limit their results to the rural population. Perhaps we are to take from that that the results only hold for rural populations overall.

At the start of this post, I noted that the contact hypothesis requires certain conditions. Schindler and Westcott come back to those conditions in their conclusion:

Taken as a whole, our results provide support for the “contact hypothesis” (Allport, 1954), which postulates that contact between groups can reduce animosity towards the minority group, and show that such effects can persist in geographies across time.

It is interesting to note that the contact which we describe meets many of the conditions that Allport postulated were necessary for intergroup contact to lead to improved relations: equal status, common goals, intergroup cooperation, and personal interaction. Black G.I.s were in the UK for a relatively short period of time, to support the war effort, and did not compete for jobs or public goods with the local population.

Of course, those conditions are not always met. For instance, in many Western countries in recent times, some migrant groups or refugee groups would struggle to meet any of the four conditions. That suggests a role for government or non-government organisations to improve the conditions of contact between these groups and the majority population.

[HT: Marginal Revolution, last year]

Monday, 13 September 2021

Mechanical clocks, the printing press, and the Reformation

Two of the most important technological innovations of the medieval period were the mechanical clock, and the printing press. The mechanical clock enabled greater coordination of people, while the printing press facilitated a rapid increase in basic literacy. However, I hadn't realised how closely linked the two innovations were, until I read this new article by Lars Boerner (Martin Luther University), Jared Rubin (Chapman University), and Battista Severgnini (Copenhagen Business School), published in the journal European Economic Review (ungated earlier version here). Boerner et al. link both technologies to the Reformation. As they explain, mechanical clocks are related to the printing press:

We argue that clocks contributed to economic and political change via two pathways, one direct and one indirect. The direct pathway was through technological agglomeration. Clocks required an immense amount of mechanical knowledge to build and operate, and their production required precision, technical skills, and dexterity in using different metals. These were precisely the type of skills that were useful for operating and repairing printing presses. It is therefore possible that spillovers from the clock’s presence encouraged adoption of the press.

And through to the Reformation:

It has long been conjectured that the press helped facilitate the spread of the Protestant Reformation, arguably the most important social, religious, and political movement in early modern Europe.

This link from the printing press to the Reformation has been established in past research, although its importance is still somewhat contested. Boerner et al. use data on 764 cities in Western Europe, with their key variables being: (1) whether the city had a mechanical clock by C.E. 1450; (2) whether the city had a printing press by C.E. 1500; and (3) whether the city was Protestant in C.E. 1530, 1560, and 1600.

Now of course there are a lot of problems with endogeneity and omitted variables in this sort of analysis. Boerner et al. deal with these problems using instrumental variables:

We address potential endogeneity and omitted variable biases by instrumenting for the presence of a clock with the town’s past experience with solar eclipses. The idea behind this instrument, which is also used by Boerner and Severgnini (2019), is based on the fact that eclipse activities stimulated the construction of astronomical tools such as astrolabes, which were the prototype of mechanical clocks. We instrument for the spread of the printing press with the town’s distance to Mainz, the birthplace of Gutenberg’s press.

Essentially, they run a three-stage regression model, where the first stage instruments for the presence of a mechanical clock using "the number of times a town experienced multiple solar eclipses over a one hundred year span between 800 to 1283". The second stage uses the instrumented variable, and instruments for the presence of a printing press using the distance from each city to Mainz (where the printing press was invented). The third stage uses the instruments to extract a causal estimate of the presence of a mechanical clock, and the presence of a printing press, on whether each city was Protestant or not.

You may be wondering about those instruments though, especially eclipses. Boerner et al. explain that:

The rationale for using eclipses as an instrument for mechanical clocks follows from two relationships: (i) the relationship between solar eclipses and astronomic instruments (astrolabes), and (ii) the relationship between astrolabes and clocks. Regarding the first connection, the observation and documentation of the course of the celestial bodies and in particular solar eclipses elicited a special fascination... Thus, the appearance of solar eclipses created curiosity to understand and predict these movements... This broad interest created a demand for the development and use of astronomic instruments to measure and predict the movement of heavenly bodies... In particular, astrolabes and in some cases astronomic water clocks were built...

The second link is that the construction of clocks was often motivated by astronomic instruments... and that the timekeeping function was stressed in European astrolabes... The fact that most early mechanical clocks were also astronomic clocks (and instruments) supports this argument further.

Then in terms of the distance to Mainz as an instrument for the printing press:

Distance to Mainz is highly correlated with the early spread of printing because the first printers were either apprentices or business associates of Gutenberg in Mainz. The secrets of the new technology – most importantly, the process used to cast movable metal type, which required a specific combination of alloys – was closely guarded among this small group for the first few decades of print... Printers also weighed cost when considering where to spread, and they therefore broadly spread out in concentric circles emanating from Mainz...

Ok, so running this three-stage analysis, Boerner et al. find that:

...towns that were early adopters of clocks also tended to be early adopters of printing, even after controlling for unobservable covariates via instrumental variables. This finding suggests that people with the elite human capital necessary to operate and repair clocks tended to agglomerate in the same cities, thus permitting spillovers when new technologies such as the printing press were introduced. Second, the printing press was positively and significantly associated with the spread of the Reformation... Third, while we cannot say definitively that the clock was statistically related or unrelated to the spread of the Reformation, a mediation analysis reveals a positive and significant indirect effect of the mechanical clock on the Reformation, indicating an important role for technological agglomeration in the spread of the Reformation.

Specifically, cities with a mechanical clock by 1450 were about 34 percentage points more likely to have a printing press by 1500, and cities with a printing press by 1500 were 33 percentage points more likely to be Protestant by 1530, and 36 percentage points more likely to be Protestant by 1600. There was some weak evidence that cities with mechanical clocks were more likely to be Protestant by 1600 (but not by 1530, so some more research on that point would be helpful in the future).

We can conclude that the mechanical clock and the printing press were two key technological innovations of the medieval period, and together they contributed significantly to one of the key social innovations of the period as well.

Sunday, 12 September 2021

Creating papers in FaKe LaTeX

Many researchers will be familiar with LaTeX, which is a software system used for laying out documents (like research articles or presentations). It is an alternative to MS Word or Powerpoint, and is widely used in mathematically-heavy fields, because it create nice-looking equations and is more aesthetically pleasing overall than most documents created in Word or Powerpoint. That means that it is somewhat of a de facto standard in economics as well (for example, follow any of the links on my blog to an NBER Working Paper, and you will see the format in action). The problem with LaTeX is that there is a steep learning curve, and if you're not writing loads of mathematical equations, it isn't clear that the benefits of learning LaTeX outweigh the costs.

However, help is at hand. This 2019 article by Scott Irwin, published in the Journal of Economic Surveys (open access), describes how. As to why you would want to do this, Irwin notes that:

This led me to think that LaTeX was just another example of a ‘cool kids club’ that could be used to exclude the ‘uncool kids’...

If you want to join the 'cool kids club' without expending the time and effort to learn LaTeX, this might be a way to fake your way in. Irwin describes step-by-step how to create a LaTeX-looking document in MS Word (and a way to create a presentation in Powerpoint that looks like it was created using Beamer, the presentation software that LaTeX uses). Alternatively, he has templates on his personal website (scroll down, and you can find them under 'Miscellaneous').

Irwin's article summarises the pros and cons of the FaKe LaTeX approach as:


  • Looks great
  • Simplicity of Word
  • Ease of sharing documents
  • Tracking changes for shared editing
  • Tables and charts may be produced in Excel
  • Quick and easy to learn and implement


  • Math formatting may not be optimal
  • Less availability of journal templates
  • Reference management not automated

Another con is that it doesn't appear that you can easily create the in-document hyperlinks to tables and figures that are a feature of most LaTeX documents. However, the templates still create LaTeX-adjacent documents that plausibly look like they were created in LaTeX to the untrained eye. So, if you too want to join the 'cool kids club', now you can!

Saturday, 11 September 2021

Performance pay, stress, and alcohol and drug use

Employers use performance pay to incentivise their workers to work harder. However, that increases workers' stress and uncertainty, both of which may in theory lead to an increase in workers' use of alcohol and drugs as a coping mechanism. In a new article published in the Journal of Population Economics (ungated earlier version here), Benjamin Artz (University of Wisconsin-Oshkosh), Colin Green (Norwegian University of Science and Technology) and John Heywood (University of Wisconsin-Milwaukee) test this theory, using data from the U.S. National Longitudinal Survey of Youth 1997 (NLSY). The NLSY collects data on whether the respondents received performance pay ("tips, commissions, bonuses, incentive pay, and a small “other” category") in each wave, as well as marijuana or alcohol use within the previous 30 days, and other drug use (which they refer to as 'hard drugs') within the last year. The sample size is over 60,000 observations.

The problem with a simple analysis here is that the types of people who are more likely to accept a job with performance pay (i.e. younger people, and people who are less risk averse) are also those who are more likely to consume alcohol and drugs. So, a positive correlation between performance pay and alcohol and drug use is to be expected because of 'selection bias'. Artz et al. use a number of techniques to get around this bias:

First, we include proxies for risk preferences, ability, and personality incorporating sophisticated error structures... Second, we use the survey’s panel structure to hold constant time-invariant worker fixed effects that could include unmeasured risk preferences or ability. Third, we recognize that changes in unmeasured worker characteristics can lead to both job change (and so performance pay receipt change) and to a change in substance use. We respond by controlling for job match fixed effects. Thus, we examine the change in individual workers’ substance use when their employer changes their performance pay status (even as they remain in the same detailed occupation).

So, they control for risk preferences (albeit imperfectly - I'll come back to that point), and then in their most restrictive specification the fixed effects reduce their comparisons to the same worker, working for the same employer. Essentially, they look at what happens when a worker goes from not having performance pay, to having performance pay (or vice versa), while still working for the same employer. In the first specification (excluding risk preferences and the fixed effects), they find that:

The main estimates of interest reveal large, positive, and statistically significant, relationships between performance pay and all types of substance use. The odds ratio... indicate that holding other determinants constant, performance pay workers have odds that are 29% higher for marijuana use, 35% higher for hard drugs use, and 45% higher for alcohol consumption.

The inclusion of risk preferences (which are measured only in the 2010 wave) do not change the estimated relationships greatly. Then, in their most restrictive specification (with worker-employer-job fixed effects), they find that:

The results indicate a 29% increase in the odds ratio for marijuana use, a 26% increase in the odds ratio of hard drug use, and a 34% increase in the odds ratio alcohol use... In sum, the results... indicate that the relationship between PRP receipt and marijuana, hard drugs, and alcohol use persists despite worker sorting on time fixed unobserved worker characteristics, or worker sorting across employers.

A potential problem with these analyses is that risk aversion decreases as people get older, and alcohol and drug use also decrease, and so even though they control for risk aversion measured at a single point in time, or control for job-match fixed effects, they still potentially don't fully eliminate the selection bias. However, it's difficult to see how this research design could be much improved upon. While jobs can be randomised to receiving performance pay, researchers couldn't force workers to accept the performance pay condition, so a randomised experiment would not work.

Performance pay does increase incentives for work effort in some jobs, but that doesn't mean that it comes without cost. Some of the cost is of course borne by the employer, but workers also must endure a more uncertain and stressful work environment. The results of this study are consistent with that story, and that some workers respond by self-medicating with alcohol and drugs.

[HT: Marginal Revolution]