All posts by Grieve Chelwa

Grieve is currently a PhD student in economics at the University of Cape Town in South Africa. Grieve has worked and lived in 5 African countries, in four of those as an immigrant. He is originally from Zambia. See our blog post introducing Grieve, or all blog posts by Grieve.

Rolf Dobelli on Citizenism

I am currently reading The Art of Thinking Clearly by Rolf Dobelli, which gives a summary of the many cognitive errors that human beings are prone to make, errors that inhibit our ability to think clearly and logically. The book also gives advice on how best to overcome some of these biases. One chapter in the book struck me as being relevant to the debate on the moral relevance of countries and the ideas around citizenism, a debate to which my co-bloggers Sebastian Nickel and Nathan Smith have recently made contributions (see here and here). The chapter is called “Why You Identify With Your Football Team: In-Group Out-Group Bias” and it appears as Chapter 79 in the book. Below I reproduce, with the author’s permission, the entire chapter (I have added hyperlinks):

When I was a child, a typical wintry Sunday looked like this: my family sat in front of the TV watching a ski race. My parents cheered for the Swiss skiers and wanted me to do the same. I didn’t understand the fuss. First, why zoom down a mountain on two planks? It makes as little sense as hopping up the mountain on one leg, while juggling three balls and stopping every 100 feet to hurl a log as far as possible. Second, how can one-hundredth of a second count as a difference? Common sense would say that if people are that close together, they are equally good skiers. Third, why should I identify with the Swiss skiers? Was I related to any one of them? I didn’t think so. I didn’t even know what they thought or read, and if I lived a few feet over the Swiss border, I would probably (have to) cheer for another team altogether.

This brings us to the question: does identifying with a group — a sports team, an ethnicity, a company, a state — represent flawed thinking?

Over thousands of years, evolution has shaped every behavioural pattern, including attraction to certain groups. In times past, group membership was vital. Fending for yourself was close to impossible. As people began to form alliances, all had to follow suit. Individuals stood no chance against collectives. Whoever rejected membership or got expelled forfeited their place not only in the group, but also in the gene pool.  No wonder we are such social animals — our ancestors were, too.

Psychologists have investigated different group effects. These can be neatly categorised under the term in-group-out-group bias. First, groups often form based on minor, even trivial, criteria. With sports affiliations, a random birthplace suffices, and in business it is where you work. To test this, the British psychologist Henri Tajfel* split strangers into groups, tossing a coin to choose who went to which group. He told the members of one group it was because they all liked a particular type of art. The results were impressive: although A) they were strangers, B) they were allocated to a group at random and C) they were far from art connoisseurs, the group members found each other more agreeable than members of other groups. Second, you perceive people outside your own group to be more similar than they actually are. This is called the out-group homogeneity bias. Stereotypes and prejudices stem from it. Have you ever noticed that, in science-fiction movies, only the humans have different cultures and the aliens do not? Third, since groups often form on the basis of common values, group members receive a disproportionate amount of support for their own views. This distortion is dangerous, especially in business: it leads to the infamous organisational blindness.

Family members helping one another out is understandable. If you share half your genes with your siblings, you are naturally interested in their well-being. But there is such a thing as ‘pseudokinship‘, which evokes the same emotions without blood relationship. Such feelings can lead to the most senseless cognitive error of all: laying down your life for a random group — also known as going to war. It is no coincidence that ‘motherland’ suggests kinship. And it’s not by chance that the goal of any military training is to forge soldiers together as ‘brothers’.

In conclusion: prejudice and aversion are biological responses to anything foreign. Identifying with a group has been a survival strategy for hundreds of thousands of years. Not any longer; identifying with a group distorts your view of the facts. Should you ever be sent to war, and don’t agree with its goals, desert.

*Henri Tajfel’s classic paper on the behaviour of groups is Experiments in Intergroup Discrimination.

The Scramble for Africa, fractionalization and open borders

Co-blogger Nathan Smith’s plea for someone to write a history of borders got me thinking about Africa and how its borders were drawn. The Berlin Conference of 1884-85 formalized what has come to be known as the “Scramble for Africa“, a process through which contemporary African borders were drawn. And as many researchers have shown, Africa’s borders were largely drawn in an arbitrary manner with little regard for the interests of the people who would later fall under the jurisdiction of those same borders. Michalopoulos and Papaioannou summarize the Scramble for Africa thus [footnotes omitted and my own emphasis]:

The key consideration of European leaders [in drawing up Africa’s borders] was to preserve the “status quo” preventing conflict among Europeans for Africa (as the memories of the European wars of the 18th-19th century were still alive). To this objective the Europeans divided areas and drew borders in maps, without taking into account local conditions and the ethnic composition of the areas. African leaders were not invited and had no say on the drawing of political boundaries. Moreover, European leaders were in such a rush that they didn’t wait for the new information arriving from explorers, geographers, and missionaries.

There is little disagreement among historians that the scramble for the continent was to a great extent artificial (see Asiwaju (1985) and Englebert (2009) for references). As the British prime minister at the time Lord Salisbury put it, “we have been engaged in drawing lines upon maps where no white man’s feet have ever tord; we have been giving away mountains and rivers and lakes to each other, only hindered by the small impediment that we never knew exactly where the mountains and rivers and lakes were.” Asiwaju (1985) summarizes that “the study of European archives supports the accidental rather than a conspiratorial theory of the marking of African boundaries.” In line with the historical evidence, Alesina, Easterly, and Matuszeski (2011) document that eighty percent of African borders follow latitudinal and longitudinal lines, more than in any other part of the world.

A direct effect of the Scramble for Africa was the partitioning of ethnic groups, many of which had existed as unitary “nation-states” for many of years. Michalopoulos and Papaioannou quantify this partitioning effect thus:

 Quantifying the effects of the Scramble for Africa requires identifying the partitioned groups. To do so we use anthropological data from the pioneering work of George Peter Murdock (1959), who has mapped the spatial distribution of 834 ethnicities at the time of colonisation in the mid/late 19th century. [We] classify as partitioned groups those ethnicities with at least 10% of their total surface area belonging to more than one country. There are 231 ethnic groups with at least 10% of their historical homeland falling into more than one country. When we use a more restrictive threshold of 20% there are 164 ethnicities partitioned across the national border.

Our procedure identifies most major partitioned ethnic groups. For example, the Maasai have been split between Kenya (62%) and Tanzania (38%), the Anyi between Ghana (58%) and the Ivory Coast (42%), and the Chewa between Mozambique (50%), Malawi (34%), and Zimbabwe (16%).  We also calculate the probability that a randomly chosen pixel of the homeland of an ethnic group falls into different countries. The ethnic groups with the highest score in this index are the Malinke, which are split into six different countries; the Ndembu, which are split between Angola, Zaire, and Zambia; and the Nukwe, which are split between Angola, Namibia, Zambia, and Botswana.

Another direct effect of the Scramble for Africa was the mishmash of different ethnic groups under a single border. The degree of such mishmash can be quantified by calculating an ethnic fractionalization index for a country, with the index ranging from zero to one. An index closer to one implies that a particular country is very ethnically diverse in the sense that the probability that two randomly picked individuals belong to the same ethnic group is very low. Using the indices calculated by Alesina et al., the average ethnic fractionalization index for sub-Saharan Africa is 0.65 with a median of 0.73. On the other hand, the average for Western Europe is 0.20 with a median of 0.12. The top 5 African countries with the highest indices are Uganda (0.93), Liberia (0.91), Madagascar (0.88), Congo DR (0.87) and the Republic of Congo (0.87). In Western Europe’s case the top 5 are Belgium (0.56), Switzerland (0.53), Luxembourg (0.53), Spain (0.42) and Germany (0.17). If the top 5 European countries were in Africa, they would rank 37th, 38th,39th, 42nd and 51st respectively in terms of ethnic fractionalization. The fact that Western European countries are not as ethnically diverse as African countries should not be surprising. The process of nation-state formation was more systematic in Western Europe with the result that the nation-state formed around less ethnically heterogeneous groups.

The effects of Africa’s arbitrary border formation have not been benign, as one would expect. In a now famous scholarly article from 1997, Easterly and Levine showed that Africa’s dismal post-independence economic performance was largely driven by the high degree of ethnic fractionalization. The authors showed that “ethnic [fractionalization] was closely associated with low schooling, underdeveloped financial systems, distorted foreign exchange markets, and insufficient infrastructure”. And this was driven by the fact that “[ethnic fractionalization] leads to rent-seeking behavior and reduces the consensus for public goods, creating long-run growth tragedies” (my italics). Even more tragic, recent work by Michalopoulos and Papaioannou shows that the random partitioning of ethnic groups explains much of the continent’s civil wars since the 1960s. Their work shows that “civil conflict intensity, as reflected in casualties and duration, is approximately 25% higher in areas where partitioned ethnicities reside (as compared to the homelands of ethnicities that have not been separated)”. The groups that have been particularly impacted by this are “the Afar and the Esa, which during the period from 1970 to 2005 have experienced five civil wars…[T]he Afar being partitioned between Ethiopia, Eritrea and Djibouti, and the Esa being split between Ethiopia and Somalia”. And the effects are not only limited to partitioned ethnic groups: “Tribal areas adjacent to the ethnic homeland of partitioned groups also experience more civil wars, which tend to last longer and be more devastating. [E]stimates imply that an ethnic group residing adjacent to a partitioned ethnic homeland is on average 5% more likely to experience civil conflict”. And civil conflicts have led to the death of millions of people in Africa and to the displacement of many more.

One can easily imagine that a different border configuration that, for instance took into account the historical interests of different ethnic groups, might have kept at a minimum some of the problems highlighted in the previous paragraph. And this is why I believe that global open borders can go a long way in rectifying some of the consequences of the Scramble for Africa. Under open borders, individuals belonging to certain ethnic groups or entire ethnic groups can easily relocate to countries that the individual or the ethnic group indentifies with or to countries that are likely to guarantee the safety of the individual or his group. (In this post, I showed that the fact that the US has open borders between states made it easier for targetted ethnic/racial groups to migrate elsewhere). Currently the costs of emigrating, even within Africa, are substantial. Very often those on the receiving end of ethnic conflicts are not allowed to easily intergrate into the neighbouring countries to which they run to but are instead placed in refugee camps known for their notoriety. Similarly, those individuals or groups seeking a better life economically in a neighbouring country are required to obtain a dossier of documents, most of which are costly to obtain, before crossing the border to begin the employment search. Those who are unable to obtain these documents resort to risky methods to get themselves across (Update: for more on this last point, see John Lee’s post titled Risking death to get into South Africa).

Some might object that whereas global open borders might lead to a reduction in ethnic diversity in Africa, for instance, they might, on the other hand, lead to an increase in ethnic diversity in certain parts of the developed world such as Western Europe or the United States. And with ethnic diversity increasing, the very problems that open borders were meant to correct in Africa might crop up in the West. This fear, however, is not borne out by research. An exhaustive literature survey by Alesina and La Ferrara finds that “rich democratic societies work well with [ethnic] diversity, in the case of the United States very well interms of growth and productivity”. This result seems to run through institutions. The presently developed countries are developed largely because their initially homogeneous populations built a consensus around establishing a set of inclusive institutions. (It is this process of consensus building that was likely short-circuited in Africa’s case during the Scramble for Africa). And these institutions are unlikely to change in the face of increased diversity because institutions, once established, tend to persist. And it goes without saying that the types of individuals or types of ethnic groups that are likely to relocate to the West are the kinds of individuals or kinds of ethnic groups that identify with the West’s way of life, including its institutions.

UPDATE: Co-bloggers Vipul Naik and Nathan Smith have previously addressed, in some great detail, the relationship between immigration, immigrants and institutions here, here and here.

Related reading

Open borders, crime and targeting the natives: the case of South Africa

Two objections to open borders are that under open borders (1) recipient countries are likely to experience an increase in crime and (2) natives, since they are on average richer than the new comers, are likely to bear the full brunt of crime. The two objections are not necessarily related: one can think of a scenario whereby the crime rate falls following open borders but the residual crime is disproportionately borne by natives. This post does not confront the first objection, namely that open borders might lead to an increase in crime, because others on this site have done so. In addition, I have speculated in a previous post that the increase in crime that immediately followed the collapse of apartheid in South Africa (to all intents and purposes an “open borders event”) was due to the fact that background checks were never performed on the “new comers” to isolate those with criminal records. (In that post, I also noted that it was impossible to perform such checks in South Africa’s case without upsetting an already delicate situation). My intention in this post is to confront the second objection, namely that natives are likely to bear the brunt of any crime that takes place under open borders. I will do so by appealing to South Africa’s post apartheid experience.

The dominant narrative on South Africa is that crime increased and has continued to increase in the post apartheid period and whites have largely been the victims of the crime wave. I took on the first part of this narrative in this post and showed that it was largely false — homicide rates (the most reliable measure of crime) have been declining every year since 1995 (apartheid officially ended in 1994). As a matter of fact, homicide rates are 50% lower today than they were in 1995. The second part of the crime narrative has a common sense appeal: whites are disproportionately the victims of crime because the average white South African is many times richer than the average black South African making him/her an attractive target for criminals. Official confirmation of this narrative seemed to have arrived when in 2009 Canada’s Immigration and Refugee Board (IRB) granted Brandon Huntley, a white South African living in Canada, refugee status on the basis that “[Huntley] would stand out like a ‘sore thumb’ due to his color in any part of [South Africa]”. Huntley “reported being the victim of several assaults by black South Africans. He alleged that those assaults were racially motivated, and stated that he did not seek police or state protection because the authorities [were] unwilling or unable to help white South Africans…The [IRB] found [Huntley] credible and accepted his evidence with regard to the attacks. One of [Huntley’s] witnesses, Ms. Kaplan, whose brother was victimized by black South Africans, also testified that the police [were] corrupt and [do] not help white South Africans, and that a genocide is occurring against white South Africans” (source). The ruling by the IRB sparked off a series of heated debates in South Africa with some seeing the decision as vindication of their long-held suspicions and others condemning it as unfounded. This prompted James Myburgh, the editor of Politicsweb, a popular local news and opinion website, to analyze several rounds of national victimization surveys whereupon he found that “whites were somewhat more likely to fall victim to crime than other race groups”. Myburgh’s conclusions were regarded as the final verdict on the matter by many since national victimization surveys were nationally representative and were conducted by Statistics South Africa, the Institute for Security Studies and the Human Sciences Research Council, bodies widely regarded as credible. The matter did not rest there, however. In December 2009, Gavin Silber and Nathan Geffen published an article in SA Crime Quarterly arguing that national victimization surveys (or NVSs) were not the best tool to use in answering the crime victimization question because they relied heavily on people’s opinions. Silber and Geffen advanced the following reasons [footnotes omitted]:

Firstly, the very concept of crime or criminality can be relatively subjective, as indeed is the case with ‘victimhood’. Some respondents to NVSs classify the threat of violence as a criminal act, while others might only classify its use as criminal. In addition, the distinction between perpetrator and victim can also be somewhat blurred in cases such as assault.

Secondly, there is strong evidence showing that reported victimization levels tend to increase with education, which is obviously (and particularly in South Africa) linked to income. A study in the United States showed that people with university degrees recalled three times as many assaults as those with a high school education. It is conceivable that over-exposure to a particular crime category amongst certain NVS respondents (in this case people with little education) may result in lesser infringements – such as assault – not qualifying as ‘criminal’. This has also been observed in studies illustrating how various developed cities/countries have produced higher victimization rates than poorer countries with higher levels of recorded crime.

Thirdly, reporting of property-related and violent crime tends to differ significantly, based on various circumstances. When a given sample is questioned on exposure to violent interpersonal crimes such as assault and sexual abuse (particularly when it involves a non-stranger), the results are likely to reflect a significant underreporting of actual exposure, due to a reluctance to report sexual abuse, child abuse, general assault and domestic violence.  Moreover, there might be differences in the way poor and relatively wealthy respondents perceive property-related crime. Relatively wealthy people have more items of value, and are able to afford insurance, which requires reporting such crimes to the authorities. This might mean they are more likely to be conscious of, or remember, thefts they have experienced in the period covered by the NVS.

They proposed an alternative method that looked at trends in deaths from non-natural causes, and in particular homicide rates. Such an approach yielded a different set of conclusions to those arrived at by Myburgh. For instance [footnotes omitted]:

In 2008/2009 18,148 people in South Africa were murdered. This amounts to 37.3 people per 100,000, or just under 50 per day. The evidence we have examined indicates that the victims are disproportionately african and coloured working class people.  We examined Statistics South Africa mortality data to determine the breakdown of murders by race. Our analysis is inconclusive but it indicates that victims are disproportionately africans and coloureds.

More compelling data come from the Medical Research Council (MRC). In an investigation into female homicide rates in South Africa in 2004, the MRC used national mortuary data to determine that 2.8 of every 100,000 white women die as a result of murder, whereas 8.9 africans and 18.3 coloureds meet the same fate. This shows that, at least for women, Myburgh is very likely wrong […] Black women are disproportionately murdered.

Another recent study by the Centre for The Study of Violence and Reconciliation (CSVR) analysed homicide rates in high-risk areas in Kwazulu-Natal, Western Cape, and Gauteng, using a representative sample of police dockets. Of the sample 85% of homicide victims were black, 9% were coloured, 5% asian and 1% of victims were white.

Silber and Geffen also looked at mortality reports produced by Statistics South Africa going all the way to the year 2000 and arrived at similar conclusions to those reported above.

So using an approach that counts the actual victims of crime shows that, contra Myburgh, the burden of crime in post-apartheid South Africa has largely been borne by the country’s non-white population.

The above notwithstanding, it is still quite possible that the homicide rate of whites has gone up since 1994 even though whites bear a lesser burden of crime when compared to other groups. Unfortunately, race specific homicide data were not collected prior to 1994 making it difficult for us to  work out and compare victimization rates for whites before and after apartheid. But the information that exists currently suggests that homicide rates for whites have come down or at the very least stayed the same since 1994. For instance, South Africa’s homicide rate began rising in the mid-1950s and continued on this upward trend, with an additional 5,000 homicides per decade, until 1994 (see the figure on page 10 of this document). And this happened during the period when the definition of South Africa excluded the homelands where most blacks (who were the majority of the population) lived suggesting that the increase in homicide rates over this period must have been largely borne by the citizens of South Africa as it was then defined. This piece of evidence coupled with the fact that homicides rates for the country as a whole have fallen by 50% after 1995 suggests that the homicide rate for whites must have declined alongside the country’s homicide rate or at the very least stayed the same since 1994.

Or perhaps certain types of victimization have increased since 1994 even though whites as a group are less vulnerable to crime when compared to whites before 1994 or when compared to other racial groups in contemporary South Africa. The typical case that is usually cited to illustrate this concern are the widely publicized attacks on white farmers. However, even here the current evidence seems to suggest that white South African farmers and their families are no more likely to fall victim to murder than the typical South African citizen. It is however important to point out that there is currently not a systematic and reliable way of collecting data on attacks on white farmers so the best analyses rely on data triangulations and assumptions implying that there is bound to be some dispersion in the conclusions. But be this as it may, all reasonable analysts of this story agree that there is currently no evidence to suggest that there is a genocide of white South African farmers, contrary to what some groups are claiming. Further, all experts agree that these attacks are not racially motivated nor do they have a political angle to them. According to a researcher who has had first-hand experience in investigating farm attacks, “most farm murders were criminal acts committed with a material motive behind them“.

What implications, if any, might this have for the open borders discussion since I have argued elsewhere that the fall of apartheid was an open borders event? Firstly, South Africa’s experience shows that crime rates do not necessarily rise following open borders — in South Africa’s case, the rate of crime has fallen in each year since 1995. And secondly, it is not clear-cut that the burden of crime will always be borne by natives following open borders — in South Africa’s case, and contrary to conventional wisdom, the victims of crime have overwhelmingly been the black (and to some extent, the coloured) population. In any case, open borders would present a way out for people such as Brandon Huntley who believed they were in danger of victimization. And under open borders, people like Huntley would not have to prove their case.

The specter of crime does not, in my mind, constitute sufficient grounds to overturn the presumption in favor of open borders — South Africa’s experience, with zero background checks, shows that crime is much more a concern for the new comers than it is for the natives. And I think it is possible to reduce the possibility of crime to a minimum by relying on background checks among other options. The alternative of consigning poor people to countries where they are perpetually the victims of crime, and to say nothing of poverty, is much worse.



Open borders and the Curley effect

James Michael Curley, a four-time mayor of Boston, used wasteful redistribution to his poor Irish constituents and incendiary rhetoric to encourage richer citizens to emigrate from Boston, thereby shaping the electorate in his favor. As a consequence, Boston stagnated, but Curley kept winning elections.

Thus begins the abstract to an interesting 2005 paper by Edward Glaeser and Andrei Schleifer that formally models what they term as the “Curley effect”. With the Curley effect, an incumbent can almost guarantee reelection by instituting policies that result in the mass exodus of that part of the electorate likely to vote against him/her (such as the “rich” or a different ethnic group). These policies are most certain to reduce the welfare (sometimes in a drastic way) of the entire jurisdiction under the incumbent’s control. Among the politicians who put the Curley effect to “good” use, in addition to James Michael Curley, were the former mayor of Detroit Coleman Young and the current president of Zimbabwe Robert Mugabe. Glaeser and Schleifer have this to say about Young:

Coleman Young was elected the first black mayor of Detroit in 1973 in a four percentage point victory over John Nichols, the white police commissioner. The election split along racial lines. Every white precinct and more than 90% of the white vote favored Nichols. Every black precinct and more than 90% of the black vote favored Young.

Young’s racial favoritism can be seen in his tax policy and his distribution of city services. A 1982 referendum tripled the commuter tax from 0.5% to 1.5%, and raised the residents’ income tax rate from 2% to 3%. This tax, which had no impact on Young’s poorer black supporters, strengthened the incentive for the better off to leave Detroit.

Did Young hurt Detroit? Did he hurt the black residents of Detroit? There is no question that Detroit was in much worse shape when Young left office than when he first entered it. Its population fell from 1.51 million in 1970 to 1.03 million in 1990, a 32% decline. The unemployment rate as a percentage of the civilian labor force rose from 10.3% in 1969 to 20.6% in 1990. The percentage of households living below the poverty line rose from 18.6% to 29.8%. Nearly all the victims of this unemployment and poverty were Young’s black supporters. Over Young’s 20 years, surely in part due to his policies, Detroit became an overwhelmingly black city mired in poverty and social problems.

And this about Robert Mugabe:

Twenty years after Mugabe took over, Zimbabwe descended into a mire of poverty, corruption, and anarchy. Mugabe himself did not appear to care for wealth, but he did single-mindedly pursue power and office. To that aim, his policies induced the whites, and others with skills and capital [such as middle and upper class black Zimbabweans], to flee. He pursued these policies knowing, and encouraging, the emigration that would follow.

Global open borders are likely to reduce the costs of emigrating (no need to obtain travel visas, apply for asylum, obtain work permits and so on) and in this way might enhance the appeal of the Curley effect to the type of autocrat described by Glaeser and Schleifer. This is likely to be so in Africa where ethnicity is often exploited by politicians. If my hypothesis about an open borders-encouraged Curley effect is true, then one might raise the followng question: might the heightened prospect of the Curley effect (and the ensuing across the board reduction in welfare) constitute a reasonable objection to global open borders?

I don’t think so. Even if open borders led autocrats to implement the Curley effect in great numbers, the same open borders would stand to counteract the negative welfare effects. Open borders would, for instance, allow a greater proportion of those targeted by the autocrat’s policies to easily relocate to some other jurisdiction than would have been the case had there been no open borders with a Curley effect.  Those leaving would  include both the well-off  who would have emigrated anyway (have specialized skills, wealth or connections in high places across the border) and those who otherwise would not have left due to the prohibitive costs associated with moving to another country (facing border patrol, obtaining work visas, risking your life at illegal crossing pointsfacing the prospect of being placed in some notorious refugee camp, etc…).  In the extreme case, everyone who was not on the winning side would relocate leaving only those deriving some benefits (however short-lived) to remain. It is reasonable to presume that some blacks must have left Detroit for greener pastures during Coleman Young’s term as mayor, if only because it is difficult to account for Detroit’s substantial population decline by appealing to white emigration only. This was probably more blacks than would have left if citizens of Detroit had been required to obtain entry visas into the United States. In the more recent case of Zimbabwe, approximately 1 million Zimbabweans had by 2008 left Zimbabwe for South Africa. The sense I get from speaking to Zimbabweans living in South Africa is that a lot more of their brethren (spouses, children, brothers, sisters, parents and friends) would have escaped Mugabe’s ruthlessness had it been easier to relocate south of the border without having to meet many unreasonable requirements to work and live in South Africa or any other country bordering Zimbabwe for that matter. So even if global open borders were to encourage the Curley effect, it is very likely that the great portion of the negative welfare effects would be counteracted by migration – in the extreme case, all negative welfare impacts would disappear since everyone affected would find it in their best interests to leave.

I however don’t think that the sort of autocrat described by Glaeser and Schleifer is really responsive to open borders in the way suggested by my hypothesis. For one thing, the Curley effect has been documented among autocrats facing relatively open borders (Curley and Coleman) to those facing relatively closed borders (Mugabe) that it is somewhat far-fetched to suggest that being a Glaeser/Schleifer-like autocrat becomes an even more attractive proposition under open borders. But what seems likely is that open borders would present a way out for those who find themselves victims of the Curley effect. 

HT David Henderson

South Africa in the open borders debate

The experience of post-apartheid South Africa is often used by supporters and opponents  of open borders as evidence for their respective positions. Those in favor of  open borders argue that the lifting of restrictions barring the free movement of black people improved the welfare of both whites and blacks. Restrictionists, on the other hand, argue that the welfare of whites and blacks has declined significantly since apartheid. For instance, the experience of South Africa after apartheid is used by both sides in the comments section to this post by Bryan Caplan  that, in part, questions the importance of the political externalities argument against open borders. I intend to do two things in this post: (1) clarify what the consensus position is regarding the welfare of blacks and whites in post-apartheid South Africa and (2) draw out the implications of South Africa’s experience for the open borders discussion.

I. Post-apartheid welfare

So what has been happening to the welfare of blacks and whites since apartheid officially ended in 1994? To answer this question, I review trends in incomes, poverty, inequality, unemployment and crime since 1994.


From an aggregate point of view, the South African economy performed relatively well from 1994 to just before the financial crisis . For instance, the rate of growth in real GDP per capita was positive and significantly so for each year from 1994 to 2008 with the exception of 1998 (scroll to page 9 of this UN report). Leibbrandt and Levinsohn work out that real GDP per capita grew by 25% over this period, which coincidentally was the average growth rate for sub-Saharan Africa, excluding South Africa, over the same time period (I computed the latter statistic by using the World Bank’s databank tool available for free here – remember to use ‘constant prices’ rather than ‘current prices’ to calculate real, inflation-adjusted growth). For comparison, real GDP per capita declined by 15% over the period 1980 to 1994 (my own calculations using data from the World Bank’s databank tool that I link to above). Since GDP per capita might be too aggregated a measure (especially for a country like South Africa with distinct racial groups), Leibbrandt and Levinsohn, in the same paper, look at trends in actual incomes as reported by South African households and find that mean household incomes grew by 15% over the period*. In terms of racial breakdowns, the white population’s incomes grew the most at 28%, followed by blacks at 26%. Coloreds and asians saw their incomes grow by 8% and 5% respectively**. No such detailed study is available for household incomes before 1994 but the consensus is that the 1980s to early 1990s were dismal times economically. One might say that the 1980s are an unfair time period to compare South Africa’s later economic performance with because of the economic sanctions in place at the time. A detailed economic analysis, however,  pointed to other structural factors, in addition to the sanctions, as being behind the economic malaise of the 80s and very early 90s.

Poverty and inequality

The consensus position is that poverty has declined across all racial groups since 1994. Leibbrandt, Wegner and Finn writing in 2011 have this to say [emphasis mine]:

The headcount ratio for two poverty lines is shown to have fallen over time, which is in line with a substantial literature (For example, Bhorat and van der Westhuizen (2009), van der Berg et al (2008)). As shown by Leibbrandt, Woolard, McEwen, Koep (2010) and Leibbrandt, Woolard, Finn, and Argent (2010) this decline in poverty is more pronounced if measures of poverty (such as the poverty gap ratio) that are sensitive to the depth of poverty are used. There is some contention over the timing of the poverty declines. For example, Hoogeveen and Özler (2006) seem to indicate an improvement in poverty levels only after 2000. However, there is no disagreement about the long‐run trend. In addition, as illustrated by Bhorat et al (2006) and Bhorat et al (2009) there is no such contention with regard to changes in non money‐metric well‐being. In all analyses, access to services, formal dwellings and private assets are shown to improve in the period from 1996 to 2001 and then on through to 2008.

Inequality on the other hand seems to have increased over all. The widely used Gini coefficient grew from 0.67 in 1993 to 0.70 in 2008 making the country one of the most unequal in the world. However, it does seem that within-racial group inequality has been driving the overall increase. In other words, inequality between racial groups (between, say, whites and blacks) is less a factor in explaining overall inequality than inequality within racial groups. As is the case in most countries, South Africa’s high inequality seems to be explained by educational differentials.


The consensus position is that the unemployment rate has been increasing since 1994. In 1995, the unemployment rate stood at about 15% whereas it currently stands at about 25%. A widely read study pointed to structural rigidities linked to poor labor market policies as being behind the persistence in unemployment post 1994.  In the absence of such rigidities, the authors contend that the labor market would have adjusted to absorb some proportion of  the large and unprecedented increase in labor supply that followed the fall of apartheid. Others, however, believe that unemployment had been trending upwards well before 1994 and the post 1994 labor policies adopted by the new African National Congress (ANC)-led government might have exacerbated the situation.


The one sore spot of post-apartheid South Africa seems to be the perceived increase in crime, especially violent crime. Crime began trending upwards as early as the 1980s with the trend line probably becoming steeper somewhere in the early to mid 1990s. Homicide rates which were as high as 65 per 100,000 population in 1995 reduced to half that figure by 2009. On the other hand, the rate of aggravated robberies increased significantly in the post-apartheid period. The Institute for Security Studies (ISS)data shows that the rate of aggravated robberies increased from about 150 per 100,000 population in 1997 to just under 300 per 100,000 population in 2004 (an almost 100% increase!) although the rate of such robberies has recently began declining. The same data also show that the incidences of rape, car jacking and murder have been declining post 1994. It is quite likely that the story around aggravated robberies has been responsible for the high crime narrative in South Africa.


II. Implications for the open borders discussion

Reading the above set of facts, one gets the sense that life in post-apartheid South Africa has not been entirely good but neither has it been the horror story that some have made it out to be. Can one then draw some implications for open borders from South Africa’s experience during the transition from apartheid to democracy (to all intents and purposes an “open borders event”)?

a) Political Externalities

With the dissolution of apartheid institutions beginning in 1990, black africans were free to live and work anywhere in the new republic. Blacks were also conferred full citizenship rights implying that they could participate in the country’s democratic process. The fact that blacks were in the majority during the transition to democracy in 1994 probably meant that the median voter was black. Parties hoping for success at the polls had to appeal to the policy preferences of the median voter who likely preferred interventionist policies to right the wrongs of the past. This is probably one of the factors behind the ANC’s victory in 1994 and the statist policies that they later on put in place. Prima facie, it seems that political externalities might have been a factor in post-apartheid South Africa. Had the new “immigrants” been denied voting rights, then perhaps the country’s economic policies might have been less intrusive. If this line of thinking is correct, then an optimal approach to open borders might be to deny, at least at the outset, voting rights to new immigrants. On the other hand, the South African case is quite different to what is most likely to happen under global open borders. For one thing, the mass “South Africanisation” that followed the dismantling of apartheid was devoid of the “selection factor” that is likely to be important under a regime of open borders. Immigrants are likely to identify with the attributes of the destination polity (why go there in the first place?) thus making political externalities a matter of less concern.

b) Crime and other undesirable outcomes

It is quite clear that the rate of crime in the immediate aftermath of apartheid was historically high by South African standards (though important to point out that crime began rising in the 1980s). Since the crime trend line became steeper in the early to mid 1990s, it becomes tempting to conclude that the “new comers” brought the crime along with them. If this conclusion is true, then the implication is not that the new comers should not have come in the first place but that background checks were never performed on the new comers to isolate those who had a criminal record and/or were likely to perform crimes in the future. (Note that it was probably impossible to perform such checks in South Africa’s unique case without upsetting an already delicate situation). It is not clear, for instance, why the US or any other country that opens up its borders cannot perform such checks. Perhaps a policy of sudden open borders (such as happened in South Africa with an incoming population many multiples that of the home population) might place strains on the verification process. In which case a policy of gradual open borders with some learning happening along the way  might be more optimal if the concerns around crime and other undesirable elements are important.


South Africa’s experience transitioning to democracy after apartheid holds some valuable lessons for the open borders discussion. I think using South Africa to completely write-off open borders is not the right lesson to draw from this story.



* The discrepancy is probably explained by the fact that GDP per capita is  sensitive to outliers (a possibility given South Africa’s high income inequality) whereas the household income data is in natural logarithms to avoid the problem of outliers.

** South Africa is officially divided into four racial groups: blacks, whites, coloreds (people of mixed-race) and asians.