Dollar General opening may be a symptom, not a cause, of negative local market conditions

Dollar stores have a bad reputation for negatively impacting local communities (see here and here, for example). Some communities have pushed back, resisting the opening of new stores. One thing that dollar stores are accused of is hurting other local businesses, but is that the case?

That is the question addressed in this new article by Amelia Biehl, Amir Ferreira Neto (both Florida Gulf Coast University), and Juan Gomez (University of Iowa), published in the journal Economic Modelling (ungated earlier version here). They focus on the opening of new Dollar General stores in Florida, using data from Dun and Bradstreet's National Establishment Time Series dataset between 1990 and 2019. Over that period, there were a lot of store openings, with the state going from 202 stores to 1014 between 2000 and 2019. That's about one store per 23,000 people in Florida by the end of the period.

Biehl et al. look at the impact on revenue, employment, and probability of firm closure, by comparing firms operating within a half-mile of a new Dollar General store opening (the 'treated firms'), with firms operating between one and 1.5 miles away (the 'control firms'), in what is termed a difference-in-differences analysis. Biehl et al. limit their focus to firms with fewer than 50 employees, as they are more likely to be impacted by the opening of Dollar General. This still results in a sample of over 7.5 million observations of over 1.1 million firms. The hypothesis is that these small firms operating in close proximity to the new Dollar General store will be affected by its opening, whereas small firms located further away will not be.

Their results are summarised in Figure 4 from the paper, which shows an event study version of the analysis:

That figure shows the impact on each variable (revenue, employment, and probability of firm closure), year-by-year, from eight years before the store opening to eight years after. Focusing on the time after the store opening (from time zero onwards, because time -1 is the year of store opening), the positive and significant effects show that employment increases, revenue increases, and the probability of firm closure increases for treated firms compared with control firms after the store opening. These results are consistent with the overall difference-in-difference results.

However, there is a slight problem here. One of the assumptions of difference-in-differences is that there are parallel trends - that treated and control firms would have followed similar paths in terms of revenue, employment, and probability of closure, if no Dollar General store had opened. Obviously, this is impossible to establish for sure, because we don't know the counterfactual (we don't know what would have happened if Dollar General had not opened). But there is one indicator that is often used to argue in favour of parallel trends, which is to look at the estimated coefficients from the time period before the treatment started. If those coefficients are statistically insignificant, it establishes that there is little evidence that the treatment and control firms were diverging before the Dollar General even opened. But for this analysis, that isn't the case. Take a look back at Figure 4 shown above. There is a clear trend in the coefficients even before Dollar General opened, and that trend appears to continue through to the period after the Dollar General opens. If anything, the coefficients at time -1 (which is the year of opening) are unusually inconsistent with the coefficients on either side of them.

The failure of this analysis to consider a violation of the parallel trends assumption should make us very cautious about taking these results at face value. At least, I don't think they should be interpreted as showing that Dollar General leads to the closure of some local businesses, with the surviving firms having higher revenues and employment (which is how Biehl et al. interpret their results). Instead, I think it possibly shows that Dollar General stores locate in areas where stores are already closing (notice in Figure 4 that probability of closure is already positive and statistically significant for treated firms compared with control firms before the Dollar General opens).

Biehl et al. go on to look at the impacts on firms by industry and by retail category. However, given that their headline results don't really show what they are purported to show, I would place even less weight on the industry-specific results. Besides which, at that stage of the analysis they are doing multiple comparisons, meaning that some of their results might show up as statistically significant simply by chance, which they haven't adjusted for.

Overall, the results from the main analysis in this article are consistent with Dollar General finding new locations where there are open storefronts (because other firms have already closed), and setting up shop in those locations. So, it may be that it isn't the Dollar General that causes firm closures, and higher revenue and employment for surviving firms, but something else like local market conditions more generally. Considering that possibility, a Dollar General opening would be a symptom of those negative local market conditions, not a cause of them.

Why specialist vape retailers may tend to locate in more socially deprived areas

When I first started studying the social impacts of alcohol outlets, one of the things my research team and I were interested in was where alcohol outlets located. We found (see here) that off-licence outlets tended to locate in areas of high deprivation in Manukau City. I've since replicated that analysis a couple of times in unpublished work, for both South Auckland and Hamilton.

I was interested to see that this new article by Robin van der Sanden (Massey University) and co-authors, published in the New Zealand Medical Journal (sorry, no ungated version online, but you can sign up for open access for free), finds very similar results for specialist vape retailers (which are defined here). They used Google Maps and Google Street View data to locate all of the specialist vape retailers across 14 Auckland suburbs, then categorised them into three types: (1) upmarket; (2) budget; and (3) 'store-within-a-store' (which are located inside or attached to convenience stores, petrol stations or liquor stores. The main results in terms of the relationship between store numbers and social deprivation are shown in Figure 1 from the paper:

This figure shows the median number of specialist vape retailers (in total and by type) by social deprivation. In their sample, stores tend to be more likely to be located in the most deprived two deciles (9-10), and least likely to be in the least deprived two deciles (1-2). Aside from that, I wouldn't draw too much from the analysis here. Because these are median counts per suburb group (not per capita or per land area), differences could reflect population size, commercial zoning, or land area rather than ‘density’. So if high deprivation suburbs also tend to have higher populations, or to be larger in area, then the apparent relationship between social deprivation and the number of specialist vape retailers is confounded. However, at the highest level there does seem to be some tendency. Van der Sanden et al. worry about this, concluding that:

The concentration of SVRs in high-deprivation suburbs in Auckland may warrant further regulatory responses that better balance the needs of predominately adults to access vaping products as a means to stop smoking with limiting vape products to young people who have never smoked...

However, Van der Sanden et al. don't really explore why specialist vape retailers may locate in areas of high deprivation. I've done quite a bit of exploration and thinking on this in relation to off-licence alcohol outlets, and I suspect that the reasons might be similar. And it doesn't require retailers to be 'targeting' high deprivation communities in some predatory business strategy. I have a few hypothesised reasons for more specialist vape retailers in more socially deprived areas can be explained with some simple economics.

First, if a prospective retailer is looking to run a retail store that maximises profits, one of the aspects that they must consider is the costs of operating the business. Ceteris paribus (all else held constant), a store with lower costs will be more profitable. Areas of high deprivation tend to have lower commercial rents, and are therefore less costly to operate, and will generate higher profits from the same revenue.

The second hypothesis is a little more complex, and involves a bit of economic geography. Each store may have a particular 'catchment area', which is the area from which its customers come to the store. In a low deprivation area, where everyone owns a car, and often commutes a fair distance for work, the catchment area for a store might be quite large. So, stores that are located close together will be in direct competition for consumers, since their 'catchment areas' will substantially overlap. In contrast, in a high deprivation area, fewer people might own cars, or they may not run reliably, or they may only be able to afford to drive them to and from work without long side-quests to buy vapes. So, the 'catchment area' for a store will be much smaller, and stores can be located closer together without being in direct competition for consumers. And so, we might expect to see more vape stores in areas of high deprivation than in areas of low deprivation, because the retailers are trying to minimise competition with other stores (although they may then need to balance a smaller catchment, which has less spending power, against the costs of operating the store).

Finally, the differences may reflect differences in demand. If vaping rates are higher in more socially deprived areas, then demand for vaping products may also be higher in those areas, and attract more vape retailers. I don't really know whether there is a social gradient in vaping, although the New Zealand Health Survey suggests that there is, with more vaping among people living in areas in the most socially deprived quintile. Of course, there is a potential reverse causation problem with the demand-side explanation, because more specialist vape retailers located in socially deprived areas might drive more vaping in those areas.

None of that is to say that having more specialist vape retailers in more socially deprived areas is a desirable outcome (especially if they do indeed drive more vaping). Van der Sanden's proposed policy response may be appropriate. However, the situation we observe could be explained by some simple economics. So if policymakers want to reduce retail availability of vaping products, they can focus on practical levers (licensing, zoning, proximity rules) without relying on arguments about predatory business practices, or vilifying store owners (both of which I have seen in the case of alcohol retailers).

The economics of maps

I have always liked maps. When I was growing up, one of my favourite books was my Rand McNally atlas. I may even still have it, tucked away with its spine held together by masking tape (after years of overuse by my primary-school-aged self). When I'm reading some fantasy novel that has a map on the inside cover, I can find myself lost in the map before even getting to read the book, and then flicking back to the map any time some new location is mentioned. Right next to my laptop while I'm writing this is a sepia-toned desk globe than, in truth, takes up too much space on the desk but will not be foregone.

Given my interest in maps, I've been planning to read this 2020 article by Abhishek Nagaraj (University of California at Berkeley) and Scott Stern (MIT), published in the Journal of Economic Perspectives (open access), for some time (like many articles that have sat in my digital to-be-read pile for a long time). Nagaraj and Stern explain the economics of maps. This isn't the economics that uses maps, such as in the field of economic geography, but two other aspects. First, they review the economic and social consequences of maps. Second, they review the economics of mapmaking. Most of the article is devoted to the latter, and that's what I want to focus on as well.

First though, what is a map? In my classes, I use maps as an example of a model - an abstraction or simplification of reality. Nagaraj and Stern note that maps are composed of two elements: (1) spatial data; and (2) a design. As they explain:

At its core, a map takes selected attributes attached to a specific positional indicator (spatial data) and pairs it with a graphical illustration or visualization (design)...

Having separated a map into its constituent elements, Nagaraj and Stern then look at the economics of spatial data, and the economics of design. On data, they note that:

...mapping data is in many respects a classical public good. Almost by definition, mapping data is non-rival insofar as the use of data for a map by any one person does not preclude its use by others; moreover, the information underlying a given database is non-excludable because copyright law does not protect the copying of factual information. While the precise expression included within a database can be protected through copyright, the underlying geographical facts reflected in the database cannot be protected.

And just like most other public goods:

The combination of non-rivalry and non-excludability of mapping data makes its production prone to private underinvestment, providing a rationale for government support. Indeed, many of the most widely used maps rely on publicly funded geospatial data, including US Geological Survey topographical maps, Census demographic information, and local land-use and zoning maps.

On the other hand:

...there are important cases where mapping data is in fact excludable, either through secrecy or contract... Mapping data that allows for excludability exhibits properties more akin to a club good than a traditional public good. Specifically, the significant fixed costs of data collection combined with relatively cheap reproducibility creates entry barriers that supports natural monopolies or oligopolistic competition. It may be efficient for only a single firm to engage in data collection and for the industry to simply license these data (under agreed-upon contractual terms) from this monopoly provider.

Now, even when spatial data is protected and excludable:

...in the absence of perfect price discrimination, private entities may only provide mapping data at a high price (relative to near-zero marginal cost), reducing efficient access. Beyond pricing, the private provision of mapping data may additionally be concentrated in locations with high demand (such as urban areas) to the exclusion of less concentrated regions.

And that all accords with what we see. There are free sources of spatial data, which are public goods supported by governments or universities, alongside proprietary spatial databases that are club goods and only available at relatively high cost (to the dismay of researchers such as me!).

Turning to map designs, Nagaraj and Stern note that:

Like data, designs are also a knowledge good in that multiple individuals can use a particular map design (and so a design is non-rival) and the degree of excludability for a given design may vary with the institutional and intellectual property environment. With that said, a striking feature of a map design is that, almost by construction, a map is created for the purpose of visual inspection, and it is much easier to copy than a database (which might be protected by secrecy or contract). One consequence of this is that there may be underinvestment in high-quality and distinct designs for a given body of geospatial data.

They use this to explain why there is a lot of competition in the provision of map designs, which is why so many maps for particular purposes look the same. As Nagaraj and Stern explain:

A potential consequence of the non-excludability of mapping data and designs is inefficient overproduction of mapping products that compete with each other. Once a given map is produced for a particular location and application (say, a city-level tourist map), copycat maps can be produced at a lower sunk cost; because demand for maps of a given quality and granularity is largely fixed, free entry based on a given map involves significant business-stealing...

Taking both spatial data and map designs together, the role of intellectual property protection is important:

On the one hand, an absence of formal intellectual property protection leads to underinvestment in mapping data and high-quality map design, but inefficient entry by copycat mapmakers. On the other hand, a high level of formal intellectual property protection can shift the basis of competition away from imitation and towards duplicative investment. For example, over the past two decades, no less than four different organizations—including Google Street View, Microsoft StreetSide, OpenStreetCam project, and TomTom—have undertaken comprehensive and qualitatively similar initiatives to gather street-level imagery and mapping coordinates for the entire US surface road system.

So that explains why there are multiple Street View clones available. The firms are over-investing in goods that are protected by intellectual property. Do we really need multiple copycats of Google Street View? Also, in terms of intellectual property protection, I found this interesting:

In addition to employing copyright, firms often invest in additional strategies to protect their intellectual property. In particular, mapmakers have devised the idea of inserting fictional “paper towns” or “trap streets” in maps... This strategy allows them to detect rivals who might copy their data (rather than collecting similar data through an original survey) and thereby protect costly investment in original data collection. Such strategies are commonly deployed by mapmakers to this day for factual data...

Does that help to explain why people have been caught out following roads that don't exist, or trying to find towns that are misplaced? I guess that 'trap streets' or 'paper towns' are a good idea on a paper map, which requires a certain amount of attention to follow, but less suitable for digital maps that people follow blindly.

Nagaraj and Stern's article opens our eyes to the economics of maps, as well as their consequences. And now, I'm going to search my garage for my beloved Rand McNally atlas. If only I had a map to guide me as to where it is hiding!

Local restrictions against unpopular retailers

When the Sale and Supply of Alcohol Act 2012 came into force, it gave local councils the ability to develop local alcohol policies (LAPs), to control the availability of alcohol in their city or district. One of the things that LAPs could do was to restrict alcohol licenses from being granted for premises that were in close proximity to 'sensitive sites' (like schools, churches, alcohol treatment providers, etc.). At the time, there was a lot of talk from some advocates about creating wide exclusion zones around these sensitive sites. I pointed out (to a number of people) that these exclusion zones wouldn't have to be very big in order to create a de facto ban on alcohol licenses entirely. Sanity prevailed, and those councils that have these sorts of restrictions in their LAPs have not made them excessively large.

Fast-forward to 2023, and the same arguments for exclusion zones are being made in relation to vape stores. For example, the Asthma and Respiratory Foundation was last year calling for a ban on retailers selling vaping products within one kilometre of a school. How feasible is that sort of control?

Steve at City Beautiful did the GIS work and reported it last year. Here's their map of all of the areas where a vape shop could set up, if you exclude all areas that are within one kilometre of a school, and exclude all areas that are not zoned commercial:

The grey areas are places where a vape shop could not be placed. The orange areas are the few places where a vape shop could be placed, in a commercial zone and more than one kilometre from a school. That's right. Almost nowhere could have a vape shop. As Steve notes in his post:

Yes, only those few tiny scraps of land shown in orange are where vape shops could go. A couple of remote industrial or business areas, sometimes with only a single site available. If there’s already a different shop there? Tough luck. The only significant area where a vape shop could actually go is right at the northern end of the city centre - and rather than being a realistic idea for every single vape user in central Auckland to come downtown every time they need to stock up, it just shows more than anything else, how desperately Auckland’s city centre lacks a school!

Urban planning is hard. Regulations that seem sensible can have unintended (as well as intended!) consequences. The problem here is that commercial zones tend to be right next to schools. If you think about your neighbourhood shops, I bet that there is a school next door to them, or just around the corner from them. Restricting unpopular retailers from being near to schools is essentially the same as restricting them from operating in most of the commercial zones in the town or city. And proponents of these regulations forget the smallest towns, of the kind that have a single school, and a single block of shops, usually on the same small stretch of main road.

There are better ways to control demerit goods than to effectively ban them. License the sellers and restrict the number of licences, tax the sale of the product, impose minimum prices, have stringent age restrictions, or do some combination of all of these things. A de facto ban, masquerading as an exclusion zone that only applies around schools, is not the way.

And besides, the location of stores selling these products becomes almost irrelevant when you consider online sales and deliveries. These products can be purchased anywhere, with very little in the way of controls. The last couple of weeks I've been conducting fieldwork looking at same-day delivery of alcohol (and that's why this blog has been pretty quiet of late). It's not quite the Wild West, but it's far from ideal at the moment. More on that in a future post.

[HT: Eric Crampton at Offsetting Behaviour]

Security guards and bank robbery displacement

Crime is an interesting topic of study. One of the interesting aspects of it is that when a deterrent to crime is put in place, crime may be reduced in the vicinity of the deterrent, but may increase elsewhere. For example, installing CCTV cameras on main streets may reduce night-time street robberies in the areas of the CCTV cameras, but only because the criminals relocate their activities to nearby streets where there are no CCTV cameras, leaving overall crime unchanged. That is referred to as 'displacement' of crime.

In an interesting application of this theory, a new article by Vikram Maheshri (University of Houston) and Giovanni Mastrobuoni (University of Torino) looked at the effect of the location of security guards on bank robberies in Italy. They use complete data on all registered banks in Italy over the period from 2000 to 2009, which includes details of 37 different security measures. They focus their attention on the hiring (or firing) of security guards, and how that affects the number of bank robberies at the hiring (or firing) bank, and at neighbouring banks (for a variety of different neighbourhood sizes ranging from 500m x 500m grid squares to 50km x 50 km grid squares).

Using this dataset, they find that:

In markets smaller than 1 km2, we find displacement effects of 1.5 to 2 percentage points to unguarded banks... Specifically, if an unguarded bank’s neighbors hires guards, the branch’s probability of being robbed will increase by roughly 20%. However, we find no statistically significant displacement effects to guarded banks... even in the smallest markets.

No surprises there. If a criminal is thinking about robbing a bank, they may prefer to find an unguarded nearby bank when they find that their original target is guarded. However, that creates a bit of a dilemma for a policymaker (or an owner of many bank branches). Should more security guards be hired, or fewer? Maheshri and Mastrobuoni simulate the effect of different policies requiring or prohibiting security guards. They show that:

In much of the country, banning guards would lead to no more than five additional robberies. However, in metropolitan areas, we might find much greater increases. For instance, Rome, Naples, Milan, and Palermo would experience more than fifty additional robberies...

If instead all banks were required by law to hire guards... the greatest reductions in robberies are concentrated in the most densely populated areas that feature the greatest number of potential targets. These include the relatively wealthy Po’ River valley in the north (which includes Milan, Turin, Bologna, and Venice) along with the major cities of Rome, Naples, Bari, and Florence...

Hiring a security guard is costly. A rational individual bank should only do so if the cost of the guard is outweighed by the benefits in terms of bank robberies prevented (measured by the losses from bank robberies that would be avoided). That is more likely in urban areas, where banks are larger and hold more cash. However, since hiring a security guard displaces bank robberies to nearby banks, an urban bank that hires a security guard creates a problem for its neighbour banks without guards. If some (many) banks have security guards, then it is better for all urban banks to have guards. So, requiring security guards in dense urban areas makes sense. On the other hand, security guards in sparsely populated rural areas make little sense, because they don't prevent many robberies and the cost of hiring them therefore outweighs the benefits. Since few banks will have security guards, then it is better for no rural banks to have guards.

The problem with this analysis is that it is based on observations of past behaviour, and that criminal behaviour may change in response to policy changes by the banks. In the data that Maheshri and Mastrobuoni use, most of the displacement in bank robberies is very local. However, if bank robbers recognise that all urban banks suddenly have security guards, but no rural banks do, then the displacement may suddenly shift from within local neighbourhoods to between urban to rural areas. This would lead to a much wider geographical displacement effect than has been observed in the past, and place the rural banks at greater risk.

[HT: Marginal Revolution, last year]

Read more:

• The optimising behaviour of Italian bank robbers

Ethnic segregation in Sao Paulo schools, and its relationship with employment and wages

Following Thursday's post about ethnic segregation and spatial inequality in Europe, I was interested to dig out this 2017 article from my to-be-read pile, by Gustavo Fernandes (Fundacao Getulio Vargas, Brazil), published in the journal World Development (sorry, I don't see an ungated version online). Fernandes used data from the 2005 School Census and the 2010 Population Census for the city of Sao Paulo, and looked at the association between segregation within public and private schools, and employment and wages for those aged 18-35. As motivation, he notes that:

The belief that Brazil has benefitted from an absence of racial and ethnic problems has been widely accepted over the last century. Brazil has often been described as a racial democracy.

Part of the motivation, then, is to debunk this 'myth'. I'm not quite sure that this counts as debunking though:

...our results show that Sao Paulo is not a city with a high degree of segregation, especially when compared to the U.S. In the city, approximately 21.29% of students would have to change schools to a new institution in order to achieve an equal composition of students by color among the entire student population of the city.

That's a fairly low level of segregation, compared not just with the U.S., but with many other countries (for instance, there's a lot of concern about segregation in the New Zealand school system). But is segregation related to inequality? Fernandes finds that, for Sao Paulo:

...segregation is correlated with the level of development in the region, which positively affects the expected returns of brancos and amarelos and negatively affects those of pretos e pardos. This result appears to be explained by the predominance of brancos and amarelos in private schools, despite the fact that most of the population of white students attends public schools. However, the effect of segregation becomes negligible when analyzing only the outcomes of students within the public school system.

The predominance of whites in private schools may be the main reason for the deep economic inequality found in Sao Paulo among races. Those schools provide a higher quality of education in comparison to public schools. They may also offer access to social networks that lead to better jobs. Both factors can exponentially increase the average income of the entire white population, resulting in large disparities between the wages of whites and the wages of pardos and pretos.

The brancos and amarelos (whites and Asians, respectively) tend to make up the majority of the class in private schools, and it is private school segregation (and not public school segregation) that is most associated with young adult employment and wages.

Ultimately, this paper demonstrates a result that is the opposite of the paper I discussed last Thursday, where greater segregation was associated with lower spatial inequality. It is impossible to reconcile the results given the wide difference in methods (not least the difference between cross-country analysis at the regional level, and small-area analysis of neighbourhoods within a single city in Brazil). However, this does demonstrate that more research on this topic is needed.

Ethnic segregation and spatial inequality

For the last few years one of my PhD students, Mohana Mondal, has been looking into ethnic segregation in Auckland (see this earlier post on some of her work). I've also maintained an interest in income inequality. So, I was really interested to read this 2017 article by Roberto Ezcurra (Universidad Publica de Navarra) and Andres Rodriguez-Pose (London School of Economics), published in the Journal of Economic Geography (ungated earlier version here), which links those two ideas. Specifically, Ezcurra and Rodriguez-Pose look at whether ethnic segregation (the concentrate of different ethnic groups within a country) matters for spatial inequality (income inequality between regions or areas of a country).

They use data on a cross-section of 71 countries where they have regional-level data on ethnic groups and region-level GDP per capita. After controlling for various factors known to affect spatial inequality such as the average size of regions, the degree of ethnic fractionalisation of the population (which is basically a measure of how many different ethnic groups there are in a country), the stage of economic development, trade openness, country size and whether a country is a transition country, they find that:

The coefficient of the index of ethnic segregation... is in all cases positive and statistically significant at the 1% level. This implies that more ethnically segregated countries have on average higher levels of spatial inequality...

This holds both for a basic regression specification, but also for an instrumental variables regression, where they attempt to demonstrate a causal effect of ethnic segregation on spatial inequality (as an instrument, they use segregation predicted using the ethnic composition of neighbouring countries). They also show that their results are robust to using alternative measures of segregation and inequality.

Ezcurra and Rodriguez-Pose then go on to investigate potential transmission channels that might explain this relationship. They find that:

...once political decentralisation and government quality are controlled for, the coefficient of the index of ethnic segregation still remains positive, but its effect on spatial inequality is statistically significant only at the 10% level... While not conclusive, these findings suggest the possibility that political decentralisation and government quality could be possible transmission channels linking ethnic segregation and spatial inequality.

The argument is that countries with more ethnic segregation are more likely to decentralise authority to their regions, which increases inequality between the regions.

This is a nice paper, but there are a couple of aspects of the research where some further work is needed. First, this research was based only on cross-sectional data. I would like to see some analysis that included a time dimension before I would conclude definitively that this relationship is causal. Second, the instrumental variables analysis seems fine on the surface, but only until you read this bit:

...the instrument used in the article predicts zero segregation for island countries...

It's pretty difficult to defend an instrument that results in such a wildly off-the-mark prediction. Certainly, you wouldn't want to predict zero ethnic segregation in New Zealand or Australia. I wouldn't expect an alternative conception of the instrument to change the results by a lot, but I think it is worth exploring. So, while this article is interesting, there is definitely more research required in this area.

 

The geography of development and the gains from relaxing migration restrictions

I just finished reading this 2018 article by Klaus Desmet (Southern Methodist University), David Nagy (Centre de Recerca en Economia Internacional), and Esteban Rossi-Hansberg (Princeton), published in the Journal of Political Economy (it appears to be open access, but just in case there is an ungated version here). The article is quite daunting because it sets up and calibrates a complex spatial model of development, which is disaggregated down to a 1° x 1° grid across the whole world. The simulation model itself is incredibly mathematical. However, if you can put the maths aside and focus on the results, what you find is interesting and insightful.

Essentially, having calibrated their simulation model, Desmet et al. look at the 'balanced growth path' of the world economy. This is the growth path "in which the geographic distribution of economic activity is constant" - in other words, where every grid cell grows at the same rate. It can take a long time for the world economy to achieve this balanced growth path, so their model runs for 600 years into the future (starting from 2000). They also run the model backwards, and can show that it does a reasonable job of replicating the pattern of population and economic development back to 1870.

Looking forward though, they base their analysis on two main scenarios: (1) holding the current pattern of migration restrictions constant; and (2) an immediate change to the free movement of people between countries and places. Both scenarios have interesting results.

In the status quo scenario, they find that:

...over time the correlation between population and productivity across countries becomes much stronger. As predicted by the theory, in the long run, high-density locations correspond to high-productivity locations...

...the high-productivity, high-density locations 600 years from now correspond to today’s low-productivity, high-density locations, mostly countries located in sub-Saharan Africa, South Asia, and East Asia. In comparison, most of today’s high-productivity, high-density locations in North America, Europe, Japan, and Australia fall behind in terms of both productivity and population.

In case you find those results surprising, Desmet et al. explain:

This productivity reversal can be understood in the following way. The high population density in some of today’s poor countries implies high future rates of innovation in those countries. Low inward migration costs and high outward ones imply that population in those countries increases, leading to greater congestion costs and worse amenities. As a result, today’s high-density, low-productivity countries end up becoming high-density, high-productivity, high-congestion, and low-amenity countries, whereas today’s high-density, high-productivity countries end up becoming medium-density, medium-productivity, low-congestion, and high-amenity countries; the United States is among them. Australia’s case is somewhat different since its low density and high inflow barriers imply that it becomes a low-productivity, high-amenity country. 

You can group New Zealand along with Australia in that paragraph. However, the takeaway message from this scenario is that the current pattern of migration restrictions, that keeps people out of high-income countries, serves to drive population density, innovation, and economic growth in the current lower-income countries to such an extent that they overtake the current high-income countries in income per capita by the end of the simulation period. That can be clearly seen from this picture (part of Figure 3 in the paper, and there are videos of the simulation model here - this is the end of Video 1B), where warmer colours represent higher levels of income per capita:

In contrast, if migration restrictions are ended, the areas that currently have high productivity attract migrants, which boosts their population, innovation, and future economic growth. So, in the free migration scenario, Desmet et al. find that:

Because today’s poor countries lose population through migration, they innovate less. As a result, and in contrast to the previous exercise, no productivity reversal occurs between the United States, India, China, and sub-Saharan Africa... Some countries, such as Venezuela, Brazil, and Mexico, start off with relatively high utility levels but relatively low productivity levels. This means they must have high amenities. Because of migration, they end up becoming some of the world’s densest and most productive countries, together with parts of Australia, Europe, and the United States.

Coastal areas also benefit greatly when migration is free. Here's the corresponding map for the free migration scenario (part of Figure 7 in the paper, and again in the videos online you can see its evolution - this is the end of Video 3B):

You can see the difference in income per capita between the two scenarios by comparing those two maps. Another interesting point is that the effect of lifting migration restrictions is immediate, with 70.3 percent of people moving immediately once the restrictions are lifted. That demonstrates how restrictive current regulations are. There are also substantial welfare gains from free migration:

In present discounted value terms, complete liberalization yields output gains of 126 percent and welfare gains of 306 percent.

While the exact numbers depend on the particular calibration of the model, I think we can safely conclude that there are huge gains in human welfare to be had from lifting migration restrictions. As Michael Clemens noted in this article in the Journal of Economic Perspectives (open access), maintaining migration restrictions leaves trillion dollar bills on the sidewalk.

The results reported in this paper should make high-income country governments seriously reconsider their immigration policies. High immigration restrictions seriously benefit China, India and maritime Asia in the long run, and are detrimental to the future economic growth of the current high-income countries. If high-income countries want to maintain their high-income status, these restrictions need to be reconsidered. However, it would be interesting to see what happens in the simulation model when one country (or a small number of countries; or rather grid cells) lifts restrictions but others do not. Perhaps that is a future exercise, but in terms of providing input to policy decisions, it seems critical.

[HT: Marginal Revolution, last year]

The economic benefit of becoming a US state

After the Mexican-American War (1846-1848), a large chunk of northern Mexico was ceded to the United States in the so-called Mexican Cession (as a result of the Treaty of Guadalupe Hidalgo). This included the modern-day states of California, Nevada, and Utah, most of Arizona, and parts of New Mexico, Colorado, and Wyoming. We could also include the state of Texas, the annexation of which laid the seeds for the war.

This raises some obvious questions. Texas obviously thought they were better off as a state of the US. Is that actually what happened? What about the other states that joined the US as part of the Mexican Cession? And what about other territories that could easily have become states, such as Cuba or Puerto Rico?

Those are the questions that this recent working paper by Robbert Maseland (University of Groningen) and Rok Spruk (University of Ljubljana) sought to address. They use state-level and country-level data on GDP per capita, and attempt to estimate the economic growth impact of a state joining the US. This paper is interesting for several reasons, which is why we discussed it in the Economics Discussion Group at Waikato this week.

First, the paper illustrates the importance of the counterfactual - what would have happened if history had been different. In this particular case, what would economic growth have been in the Mexican Cession states, if they had remained part of Mexico? We don't know for sure of course, because we are only able to observe what happened after they became states, and we don't observe what would have happened if they didn't. Similarly, for the territories that didn't become US states that Maseland and Spruk look at, we don't observe what would have happened if they did become states, only what happened when they didn't.

Maseland and Spruk address the counterfactual problem by using the synthetic control method. Essentially, they create a model of each Mexican Cession state's observed variables (including GDP per capita) up to the time before the Mexican Cession, based on the observed variables of other countries. This creates, for each state, a synthetic version of the state that is made up of proportions of the data from other countries. For example, they find that:

...the growth pattern of prestatehood California is best reproduced by growth and development covariate values of Canada (49%), South Africa (19%), New Zealand (15%), Egypt (10%), Norway (5%), and United Kingdom (3%), respectively. On the other hand, the synthetic control group for Arizona consists of Egypt (70%), Jamaica (18%), United Kingdom (11%), and Greece (1%), respectively.

As an interesting aside, New Zealand was part of the synthetic versions of California (15%), Colorado (11%), Nevada (58%!), and New Mexico (1%).

The second interesting aspect of the paper is the results themselves. Maseland and Spruk compare the actual economic growth performance of each state with the growth performance of the synthetic version of itself. They find that there are:

...strong and pervasive effects of the admission on long-run growth. The underlying post-admission gap coefficients are both large, positive and statistically significant at 1%, respectively, and readily suggest that the effect of joining the US appears to be specific to the treated states.

For example, here's a graph of the economic growth path for actual and synthetic California:

The solid line is the actual trajectory for California, while the dashed line is the synthetic control. It is clear that they deviate from each other around 1850, and the synthetic control does much worse. This demonstrates how much better off California was after statehood.

Maseland and Spruk then go on to show that the opposite is true of Mexican states that did not join the US, finding that there are:

...large gains from the hypothetical admission of Mexican states to the United States although the variation in the long-run growth effect is notable... the states next to the US border appear to be most adversely affected by not joining the United States.

Maseland and Spruk then look at the hypothetical case of several territories joining the US, including Cuba, Puerto Rico, and the Philippines, (all of which were occupied by the US after the Spanish-American War in 1898), and Greenland (which was granted home rule by Denmark in 1979, and could hypothetically have joined the US as a state then). They find that:

By 2015, the synthetic Cuba as a hypothetical US states would be seven times richer than its real counterpart... By 2015, the difference between the synthetic Philippines as a US state, and the real Philippines is about a factor [of] 10.

...synthetic Puerto Rico as a US state would be 51 percent richer than the real non-state Puerto Rico... In terms of magnitude, the gap between synthetic Greenland as a US state and its real version as a Danish territory is 32 percent, respectively.

Finally, they look into the reasons for these substantial differences. Does statehood allow easy access to a larger domestic market and therefore drive economic growth, or does statehood lead to the adoption of better institutions? Maseland and Spruk find evidence in favour of the role of institutions:

On balance, our estimates indicate that the joint temporal and spatial variation in the level of electoral democracy can explain up to 41 percent of the statehood-induced growth premium. The corresponding variation in the level of liberal democracy accounts for up to 56 percent of the long-run growth benefits stemming from improved institutional quality upon the admission to the United States. The institutional quality bonus received by joining the US apparently drives a large part of the performance boost. This lends support to the second leg of the American Exceptionalism thesis, which is that it is America's political institutions that gave the US its unique advantage.

So, institutional quality was key to the improved economic growth performance of the Mexican Cession states after they joined the US.

There is a third reason why this paper is interesting. The original Constitution of Australia provided an opportunity for New Zealand to become a state of Australia. We chose not to. On the other hand, Western Australia became a state at that time following a referendum, when instead they could have become independent. An interesting exercise for an enterprising honours student might be to look at these two cases (data permitting) and follow the process outlined by Maseland and Spruk to construct the relevant counterfactuals for New Zealand and Western Australia. Was New Zealand better off going it alone?

[HT: Marginal Revolution]

You won't find meth labs in places where you're not looking for them

I just read this 2018 article by Ashley Howell, David Newcombe, and Daniel Exeter (all University of Auckland), published in the journal Policing (gated, but there is a summary of some of it here). The authors report on the locations of clandestine methamphetamine labs in New Zealand, based on data from police seizures between 2004 and 2009.

It's an interesting dataset and paper, and they report that:

In the unadjusted spatial scan, there were five locations in the study area with significantly high clandestine methamphetamine laboratory rates (Fig. 2). The ‘most likely’ cluster, centred in Helensville (north-west of Auckland), had a RR of 4.14 with 59 observed CLRT incidents compared to 15.1 expected incidents. A similarly high cluster (RR = 4.09, P = 0.000) was found in the Far North TA.

In other words, there were four times as many lab seizures in Helensville and the Far North than would be expected, if lab seizures were randomly distributed everywhere. The other locations were Hamilton, West Auckland, Central Auckland, and there was a sixth cluster centred on Papakura in some of their analyses. This bit also caught my eye (emphasis mine):

In addition, 26 laboratories (2%) were found at storage units, 21 (2%) discovered in motel or hotel rooms, and another 27 were abandoned in public areas, including cemeteries, parks, roadsides to school yard dumpsters and even the parking lot of a police station.

I wonder how much effort it took for police to find that last one? The paper gives some insights into where the most meth labs have been seized by police. However, we should be cautious about over-interpreting the results, because by definition, you can only seize labs in locations where you are looking for them. So, if police are more diligent or exert more effort in searching for meth labs in Hamilton or the Far North, we would expect to see more lab seizures there, even if there were actually fewer labs than in other locations.

To be fair, the authors are aware of this, and in the Discussion section they note that:

Reports of clandestine laboratory seizures may also be prone to subjectivity. There is no way to be certain that CLRT incident density is not a symptom of a greater police presence or different policing priorities.

However, that doesn't stop them from noting in the abstract that:

Identifying territorial authorities with more clandestine laboratories than expected may facilitate community policing and public health interventions.

It is true that identifying areas with more meth labs than expected would give information about resource allocation. The problem is that this paper doesn't tell us where meth labs are, it only tells us where police have found them.

[HT: The inimitable Bill Cochrane]

The spread of Christianity in Austronesian societies

What explains the success of Christianity as a religion, especially in (relatively) modern times? For example, modern Pacific Island cultures are predominantly Christian, and none of them would have begun converting until the first missionaries arrived in the 17th Century. An article last year by Joseph Watts (Max Planck Institute for the Science of Human History and University of Oxford) and co-authors, published in the journal Nature Human Behaviour (sorry I don't see an ungated version), looks into this question.

Watts et al. used data on Austronesian cultures (which are spread from Tahiti in the east to Sumatra in the west, plus Madagascar) and the timing of the conversion of half of each culture to Christianity over the period from 1668 to 1950, to test three hypotheses:

1. "whether cultures with greater political organization are faster to convert to Christianity, as predicted by top-down theories of conversion" - more politically complex societies are also more inter-connected, and the theory is that an innovation such as Christianity will spread faster;
2. "whether cultures with higher levels of social inequality are faster to convert" - more unequal societies have a social stratification, and Christianity brings a more egalitarian ideal that might appeal to those at the 'bottom'; and
3. "whether larger populations are slower to convert" - with larger populations, it simply takes longer for innovations to be adopted, particular when an innovation (such as Christianity) requires interaction between people (conversion).

They found that:

...population size was found to be significantly positively correlated with conversion times, indicating that larger populations took longer to convert to Christianity. Consistent with the top-down theory, political complexity was negatively associated with conversion times... Counter to the bottom-up theories, there was no reliable support for an association between conversion time and social inequality...

You might be wondering why I've posted about this particular research paper. It isn't because of the theory, or the results. Instead, I found the methods quite interesting. Health warning: the following description might be a little too technical for some readers.

A particular problem emerges in regression models when observations are not independent. For instance, regions that are close together spatially (e.g. neighbouring regions) are likely to be similar, and demonstrate similar relationships between variables. Treating the regions as independent observations (when they aren't, because they are neighbouring and therefore similar) leads the estimated standard errors from a regression model to be too small. The consequence of that is that we are more likely to have models tell us that coefficients are statistically significant, than they would be if we correctly treated the observations as not being independent. To deal with this in the case of regions, there are spatial econometric models.

Watts et al. don't deal with the problem of spatial dependence, but they do deal with a closely related problem that I have been thinking about for the last year or more - cultural dependence. We often run cross-country regression models (e.g. for happiness studies), treating the country-level observations as independent. However, countries that have similar cultures are not independent observations, and we should be accounting for that. Watts et al. do this in their model:

Standard regression methods assume that cultures are independent from one another, despite them being related through common descent and patterns of borrowing... This non-independence can lead to spurious correlations, and the difficulty of distinguishing such spurious correlations from correlations that result from actual causal relationships between variables has come to be known as Galton’s problem. The PGLS-spatial method developed by Freckleton and Jetz makes it possible to address Galton’s problem using a phylogeny to control for non-independence due to common ancestry, and geographic proximity to control for non-independence due to diffusion between cultures.

Their phylogenetic approach uses a language-based family tree to define how close (or far away) each Austronesian culture is from others. This is a nice approach to dealing with the issue of cultural dependence between the observations, and something we should make more use of in regression models. The irony is that in the case of this paper, the phylogenetic and spatial dependencies turned out to be statistically insignificant. I guess that, sometimes, a lack of independence between observations isn't as big a problem as we may worry it is.

[HT: New Zealand Herald, last year]

What to do when your cities are stuck in the wrong place

The persistence of the location of cities and towns is well recognised in economic geography. People tend to locate where jobs are. New industries (and hence jobs) tend to locate close to where customers are, which unsurprisingly, is where people are. And so, the location of cities in the future is likely to be where cities were in the past.

In order to get substantial change in the location of towns and cities, it looks like you need to generate a collapse in civilization. At least, that might be one tongue-in-cheek take-away from a recent paper by Guy Michaels (London School of Economics) and Ferdinand Rauch (University of Oxford). Michaels and Rauch studied a cool natural experiment - the effect of the fall of the Roman Empire on the location of towns in Britain and France. The key point that makes this natural experiment useful is that the effect of the fall of Rome was much bigger in Britain than in France:

Roman Britain suffered invasions, usurpations, and reprisals against its elite. Around 410CE, when Rome itself was first sacked, Roman Britain's last remaining legions, which had maintained order and security, departed permanently. Consequently, Roman Britain's political, social, and economic order collapsed. From 450-600CE, its towns no longer functioned. The Roman towns in France also suffered when the western Roman Empire fell, but many of them survived and were taken over by the Franks.

• In short, the urban network in Britain effectively ended with the fall of the western Roman Empire; French towns experienced greater continuity.
• The divergent paths of British and French urban networks allow us to study the spatial consequences of the resetting of an urban network, as towns across Western Europe re-emerged and grew during the Middle Ages.

They find that the location of towns changed in Britain, but remained the same in France. But did that even matter? It turns out it did:

The conclusion we draw is that many French towns were stuck in the wrong places for many centuries. They could not take advantage of the new transportation technologies since they had poor coastal access; they were in locations that were designed to fit with the demands of Roman times and not the considerations of the Middle Ages.

So, towns and cities can be stuck in the 'wrong' (from a productivity perspective) location for centuries or longer. There are no barbarian invasions in our near-term future, so we are to a large extent stuck with the urban locations we have now. This has interesting implications for adaptation to climate change. Many cities currently sit in extremely vulnerable locations, in terms of surface flooding, sea level rise, desertification and water stress, etc. The implications of this paper is that there is substantial inertia that will prevent large-scale relocation of people and industries to areas that are more resilient or less vulnerable. In other words, adaptation to climate change in situ is going to be very important - we can't simply rely on moving away from where the problems occur. On a related note, we shouldn't expect large masses of migrants trying to get away from vulnerable cities and countries to suddenly end up on our doorstep. It simply isn't that easy for them to move.

For the full paper (gated), see here.

[HT: Paul Krugman's NY times blog]