The military draft, mortality and education

The data sometimes work in mysterious ways and provide puzzling correlations that lead to interesting research questions. One such correlation is that exemption from military service leads to lower mortality later in life.

Piero Cipollone and Alfonso Rosolia find this while looking at a natural experiment following the 1981 earthquake in Southern Italy. Boys from the affected region were exempted from military service, and they were followed, along with non-exempted neighbors, to track their life and education. By concentrating on boys both sides close to the border of the exempt region, they find that those exempt ended up being more educated. I can easily believe that, as they were not spending some of their prime learning years hiding in bushes and peeling potatoes, and they were expecting a longer work life. But the exempt also have lower mortality. This is not due to a lower incidence of military accidents, it is rather linked to the higher school completion rates. In fact, the authors conclude that raising high school completion by 10 percentage points would lower mortality by one or two percentage points in the decade thereafter. That is impressive at that age.

The child quality/quantity trade-off in the Industrial Revolution

Non-economists cringe when they hear us talking about investment in children and the quantity/quality trade-off in this regard. Yet, this is a very real aspect of child rearing pointed out by Gary Becker that is at the core of many models, and has been found wild in nature. This trade-off is though to be an integral part of the demographic transition, where fertility suddenly drops massively in the course of development.

Marc Klemp and Jacob Weisdorf look at data from Anglican parish registers from the 18th century that contain all sort of demographic data to look at the child quality/quantity trade-off during the Industrial Revolution. Theory tells us that if the returns to education and/or the cost of time (wages) get larger, parents switch from having many children with no education to few of them with better education. Klemp and Weisdorf's data indicates clearly that this trade-off is present: each additional sibling reduces by 8% the probability of a child eventually becoming literate. That is a strong effect, in particular considering the larger number of children at the time, and its rather large standard deviation during this time of transition.

Fertility differences and agricultural techniques

There are times when you read a paper and you really wonder how the authors came up with the idea to check out a particular correlation in the data, because it seems to be so far-fetched. But thus a correlation can be beautiful if it also has a nice theory that comes with it.

The correlation that Alberto Alesina, Paola Giuliano and Nathan Nunn study is between current fertility and adoption of plough agriculture in history. OK, I did not think about that one. But now that they find a nice positive correlation, how could one explain it? They argue that this has to do that women and children are not particularly useful when ploughing, as strength is required. The traditional task of weeding, that fell on women and children, is not necessary with ploughing. Thus, there is a preference for fewer children that is ingrained in the culture of these regions to this day.

Fertility and self-control

The Beckerian theory of fertility decisions in the family is based on the rationality and self-control of the involved parties, which is in stark contrast of the Malthusian theory of population growth, which relies on people breeding without control. As so often, the truth lies somewhere in between, as it is clear people guide their fertility outcomes but their is a substantial stochastic element to it.

Bertrand Wigniolle reinterprets Becker's theory by adding lack of self-control in the for of hyperbolic discounting, that is, discounted values fall rapidly in the near future and more slowly for the distant future. Wigniolle uses a three period model with parents valuing number and quality of children. In period 1, they choose the number of children and pay time costs for rearing them. In period 2, they choose their education, and support the associated costs. In period 3, they only enjoy their children. Hyperbolic discounting implies that every period they regret some of their past choices. The model yields very different results depending on parameter values. If parents have grounds to invest in the education of their children, then fertility is lower due to the lack of self-control. Call this a developed economy. If parents have no reason to invest in education (say because its return is low), then fertility gets higher with hyperbolic discounting. Call this a developing economy. And this is very sensitive to parameter values. A further reason to push for more schools and opportunities to use human capital in developing economies.

Rent seeking in divorce

Divorce is breaking a marriage, and courts rule on the compensation of the involved parties. In countries where no-fault divorce is allowed, how compensation is allocated is very much dependent on the outside option of each party, particularly when negotiations happen without the involvement of a judge, who is then just a threat point. For example, when there is only consensual divorce allowed, the partner not seeking divorce has all the bargaining power. But when no-fault unilateral divorce is allowed, the roles are completely reversed.

Sietse Bracke, Koen Schoors and Gerd Verschelden study how the introduction of unilateral divorce changes outcomes in Belgium, where consensual divorce was already permitted. In particular they look at self-sacrifice, for example how some household member may specialize in home production and thus jeopardize her labor market potential and bargaining position in case of divorce, especially when there is no-fault unilateral divorce. Indeed, one can view this specialization as an investment in future rents from marriage, and divorce annihilates those.

To analyze this, the authors collected survey data from divorces in four cities for a year. Using alimony as a signal of bargaining power, they find that alimonies are higher or more likely for long marriages, for no-fault divorces, and when there is significant self-sacrifice. That would all be consistent by theory, but unfortunately these results are somewhat tainted by the fact that the law gives judges very similar directives for handling divorces outcomes.

Breastfeeding and cognitive skills

Breastfeeding is now almost universally promoted as the healthiest way to feed a baby. And indeed, while breastfed babies are a little smaller and than bottle-fed ones and gain a little less weight, they are healthier, it is thought mainly because the mother milk transmits antibodies and relevant nutrients. But not every mother breast feeds, maybe because not every mother realizes all the benefits, or because some of the costs are high (time management for working mothers or aesthetic issues). Or there are some other benefits that are not well known.

Maria Iacovou and Almudena Sevilla-Sanz report that breastfeeding has significant positive impacts on cognitive skills (reading, writing and mathematics). While this correlation is well known, it may be spurious because mothers who breastfeed are more likely to be well educated (Irish example), and their children are also more likely to be well educated as well. The obvious way to overcome this statistical issue, a randomized trial, is not feasible on ethical grounds. What Iacovou and Sevilla-Sanz do is use propensity score matching, which essentially matches babies that have the same characteristics but breastfeeding and then compare their cognitive skills. What is particularly impressive in this study is that the retained characteristics are very broad beyond baby demographics and health, including parent characteristics such as education, job, income, and even pre-birth attitude towards breastfeeding or home and neighborhood. And even after controlling for all these variables, the impact of breastfeeding is still significant on babies from Bristol (England), and it may even grow with age.

Syphilis cannot be eradicated

Syphilis is back. As the most widespread venereal disease in the 1930's it was curtailed after huge efforts, both in developed and developing countries. And ten years ago a push was made to finally eradicate it. But syphilis is quietly making a comeback, hidden in the shadows of AIDS. And because the transmission of this disease is primarily driven by risky sexual behavior, it can be a leading indicator of other sexually transmitted diseases on the rise.

David Aadland, David Finnoff and Kevin Huang use a model of human behavioral response to study this resurgence and come to the conclusion that there is a fundamental cycle that cannot be broken. The point is that the transmission model that epidemiologists use has constant parameters tracing back to biological features of the disease, but these parameters can change through human intervention, and they do. Call this a Lucas Critique of epidemiology. The key here is that when prevalence is low, individuals in a riskier fashion and in particular have more sexual partners. No reasonable policy can overcome this.

Why is the Chinese savings rate so high?

The current global imbalances, at least those between the US and China, are only possible because China is currently saving a historically high share of its income. Various theories have been advanced to explain this surge in the savings rate: 1) Economic reform has increased household-level uncertainty and thus precautionary savings. 2) Forces have shifted from consumption-oriented households to savings oriented businesses. 3) Demographics and the life-cycle combined with the growth in income lead currently to high savings rates because savings change through the life cycle and thus fluctuations in the dependency ratio become important.

As Carl Bonham and Calla Wiemer point out, the savings rate has not been uniformly high and is in fact consistent with the changes in the dependency ratio. The savings rate increased through the 1980s to peak at 41.9% in 1995, then "bottomed" at 37.7.% in 2000, before surging back to 51.4% in 2008. The current global imbalance occurs in part because, unlike before, investment rates are restricted by policy, and stand at 43.5%. A modest decrease in the savings rate can rebalance things.

To test the three theories against these staggering numbers, Bonham and Wiemer use a structural VAR and determine the latter one is the most important, while the others cannot be dismissed. I am not particularly fond of VARs to test theories, they should rather just describe the data, but the evidence is quite compelling in this case. Of particular interest is that one can forecast the savings rate, as the dependency ratio is quite predictable. And this forecast shows that Chinese savings rates have peaked last year and will decrase quite significantly over the next decade. If true, this should reduce considerably the pressure on China to do something about current imbalances.

On the consequences of slavery

I reported now long ago on the consequences of slavery in Africa, where it is shown that countries where more slaves were taken still have lower levels of development. It is quite amazing that this can have an impact on average income for so long. But what about the receiving end of the slave trade?

Graziella Bertocchi and Arcangelo Dimico look at county data for the US and find that current average incomes are not related to the inflow of slaves. However, income inequality is. Why would that be? It could be because slave could not own land, and this still has an impact today. Or it could be because of discrimination. Or it could be because of persistent differences in human capital.

Now using a panel data set, Bertocchi and Dimico find the last one is the most likely one. The education gap between blacks and whites has never recuperated, and segregation was certainly part of it. But this strongly persistent effect means that affirmative action still has a reason to be.

Financial development and fertility

Why an economy's financial development matter for its fertility? For one, if there is little in terms of savings technology, then households need to find other ways in which they can save for old age, and children have traditionally been a good way to do this. But as long as property rights are reasonably well established, this should be that important, as there are many ways beyond financial assets to accumulate wealth, such as land, real estate, jewelry and cattle. Where a financial system can really bring change is when it gives access to credit for households.

Valerio Filoso and Erasmo Papagni study this with a life-cycle model where there is altruism from parents to offspring and vice-versa. They show that all depends on whether children are an inferior or a normal good, like in poor respectively rich countries. In addition, a relaxation of borrowing constraints allows greater investment in children. There is therefore ambiguity, which is compounded by price and second order effects. To sort it all out, Filoso and Papagni use cross-country data to estimate the sum of all these effects and find indeed that financial development decreases notably fertility in poor economies and increases it in rich ones. This may be an explanation for a fact that puzzled me for some time, why the US has a higher fertility than other rich countries.

How to compensate the short-lived

It is very unfortunate if you die early, and knowing this it would be optimal to compensate you for this misfortune. But this is difficult to achieve as your death is not predictable, and compensation after death is not very helpful to the deceased.

Marc Fleurbaey, Marie-Louise Leroux and Grégory Ponthière have figured out a scheme that would achieve this. First, they need to define a social welfare function that would make it desirable to compensate people for shorter lives. Second, they find a policy that would achieve such a compensation. The policy is to essentially cancel social security while putting the mandatory retirement earlier.

Basically it is all about getting people to consume early, so that the different between being dead or alive is not that large later in life. This seems very difficult to achieve given that life valuation studies indicate that people value life at a multiple of consumption. Furthermore, if we discourage people from saving that much, this must have large negative consequences for the accumulation of capital in a macroeconomic sense, something that has been neglected here, and should not.

Was Malthus wrong about mortality?

One of the critical assumptions in the Malthusian model of growth that the mortality rate depends inversely on the standard of living. While this is not that obvious to replicate with data we have from Malthus' times, such a relationship ship is even more difficult to establish for previous centuries.

Morgan Kelly and Cormac Ó Gráda have dug up some inheritance records from England which allowed them to link wealth and age of death. What they find is quite interesting: all strata of society where affected by agricultural conditions. So if there were some poor crops, rich and poor were more likely to die. This fits right into the Malthusian model where the aggregate standard of living matters. The only subtlety is that the rich die a little later. Indeed, the authors do not attribute this mortality to hunger, but rather vagrancy following some poor crop that spreads epidemics. But such a relationship seems to have disappeared by the 17th century, which the authors justify by government intervention to help the poor. This makes it unclear why Malthus made it such a crucial component of his theory.

An unexpected consequence of crises: birth weight loss

Periods of crisis generate hardship, in particular if people do not have good ways to insure against these kind of shocks. In theory, temporary losses in income should not have much of a permanent impact, but in practices they may. For example, losing a job entails a wage loss in the next job, because one may have lost human capital, firm specific skills or good outside options. And those wage losses are quite persistent.

Carlos Bozzoli and Climent Quintana-Domeque identify a different persistent effect of a crisis by looking at Argentina. They notice that babies born around 2002 have had a birth weight 30 grams lower than usual, a non-trivial difference that corresponds to a difference between the US and Pakistan. Now, birth weight has been identified to be a remarkably good predictor of future outcomes for a population in terms of health, education and income. Thus, the effects of the 2002 crisis could linger in Argentina for decades.

Why should marriages be eternal contracts?

Risk sharing contracts are good contracts, and they can have an important welfare benefit. However, it is not quite clear that these contracts should be valid for a life-time. Indeed, this is what marriage is. Aside from the fact that these contracts are restricted to be signed between people of opposite gender, one can also question why not more people could be included in them. That would be good risk diversification. The problem is that a marriage contract is not just about risk sharing (and for that we have a market for insurance contracts nowadays), it is also about limiting sex choices and commitments to supporting children. But all this could be achieved without marriage.

Anyway, Stefania Marcassa and Grégory Ponthière ask whether marriage contracts should be eternal and find that in most cases they should not. Their argument has nothing to do with risk and uncertainty, rather with declining match quality and unequal bargaining power. Indeed, when one potential partner has little bargaining power, marriage gives an important veto power. But when bargaining powers are rather equal, like they are nowadays, long-term contracts become more interesting because no one has an incentive to veto a marriage due to unequal power.

The origin of the demographic transition

Compared to two centuries ago, today's world is much different, as the standard of living has dramatically increased, along with population. This has been in strong contrast with previous history, characterized by growth close to zero in both population and the standard of living. During this period, there has been a very strong demographic change, called Demographic Transition, with a large decrease in mortality followed by a decrease in fertility. This has implied that every country that went (or still goes) through this transition has a period of high population growth while mortality is low and fertility has not yet declined. Such major shift in demographics have large implications, but it is also important to understand what triggered the Demographic Transition, especially as some countries are now just at the start of it.

Oded Galor tries to disentangle to various triggers that have been proposed. As this is a dynamic process, obviously some triggers are going to be more important at different stages of the transition. There is too discussion in the paper about the various theories and the quantitative evidence for and against them for me to summarize it efficiently here. Galor concludes that the following theories hold water when plunged into the data: First there is the theory that the higher demand for human capital during industrialization lead to a decline in fertility as parents concentrated more on the quality of their children rather than their quantity. Second, as the wage gap between females and males decreased, the increase in female labor force participation and the associated higher opportunity cost of having children for mothers reinforced the decrease in fertility.

What is important here are the theories that did not passed the test of the data according to Galor: the theory that the emergence of financial markets made in less necessary for parents to have been adult children to support them in old age; the theory that a decline in mortality lead to a too high number of surviving children; and the theory that the general increase in income lead to a rising opportunity cost of raising children.

Men last longer

Women live longer, yet paradoxically they can claim pension benefits earlier in many countries, where there is strong resistance to equalizing the retirement age (let alone increasing it, see last week's post). Would true believers in markets and efficient politics still find an explanation of this paradox?

Wolfgang Maennig and Michael Stobernack offer one: the physical performance of men declines much slower with age than for women. They base this on the worldwide top performers by age on rowing machines. The latter are of uniform quality, thus environmental factors do not matter and everyone competes on level ground. This is better than previous studies relying on track-and-field records, that more susceptible to whether influences (and doping). From the 40's to the 60's, the physical performance of men declines by about 15%. This is less than the productivity decrease that would be implied from wage changes (and labor productivity does not depend solely on physical performance, one could think older workers have in fact better non-physical qualities like experience). For women, this is more in the order of 20%. The difference is even more pronounced for those in the "lightweight" category.

The study does not go beyond the 70's (and I suppose the records pertain to the younger ones among those). So it must be that at some point the men start declining really fast. It also be that the age differences among the best rowers are simply not representative of the age differences among the general population. These elite athletes maintain their body, whereas the general population may not, and especially there may strong gender differences in doing so.

Homosexuals and savings

Do homosexuals save more or less than heterosexuals. As gays typically do not have children, one may think that they care less about the future and that this leads to lower savings. This is a prevalent intuition that even brought Hans-Herrmann Hoppe trouble over academic freedom with his administration. But is this idea true?

Brighita Negrusa and Sonia Oreffice claim it is true based on an analysis of the 2000 US census, showing that homosexuals couples accumulate significantly more retirement and social security savings. They save more than heterosexual co-habiting couples, who themselves save more than heterosexual married couples. The same can be concluded from mortgage to house value ratios.

Note that savings are not directly measured in those datasets. retirement and social security savings are inferred from retirement and social security income for retirees, and household wealth is inferred from the share of mortgage payments to the house value. And homosexual couples are also inferred by have to same gender adults in the same household. But even with all the measurement errors involved, the results are pretty clear and should at least give us to rethink some unfounded intuitions.

So why would homosexuals save more? The fact that they have fewer children, which would have accounted for the lack of altruistic bequests, may also mean that they cannot count on support from descendants. This is consistent with unmarried couples (who have less children) saving more than married couples. But why then do homosexual (unmarried, this is the year 2000) couples save more than unmarried heterosexual couples? The authors rule out discrimination on credit markets. An open question.

Women's rights and development

It is sometimes puzzling when someone in power relinquishes it, especially when this person can take advantage of institutions to stay in power. One such case is when males, who were the only ones to have property rights, gave property rights to females. Why would they allow this, given that males had monopoly on votes and government?

Raquel Fernández comes up with an elegant story. The important trade-off is between appropriating all rights to oneself and how descendants, who include daughters, get. Looking at history, there are two major effects. The first is a decline of fertility, which under a patriarchal system allows fathers to bequeath more per son. But this increases the disparity between sons and daughters, and thus reduces the advantage over equal distribution, through the concavity of utility. The second effect comes from the general increase in wealth, with a similar logic.

In both cases the father becomes better off sacrificing some wealth to his wife in order to force sons-in-law to be more generous with his daughters. The prediction is then that lower fertility and more advanced economic development both force earlier adoption of female property rights, something Fernández finds across US states, which have adopted such laws at various times between 1846 and 1920 (and four states still missing by then). And the theoretical prediction is obtained without assuming changes in altruism towards children, which I find remarkable.

Did modern growth kill inheritances?

There is a literature that tries to understand bequests, in particular whether they are accidental or voluntary. This is an important question as it has strong implications on savings motives, and thus on aggregate savings. With the development and sophistication of financial markets, one should see the accidental bequest to become less important. People can insure themselves against uncertain life times with annuities and life insurance, thus reducing the need to have excess savings in the case of unexpectedly long life. Voluntary bequests may go either way with development, depending on whether parents prefer to make inter-vivos transfers, which they can afford given the above.

Now all this is put into question after a great data digging initiative byThomas Piketty, who compiled bequest data for France from 1850 to 2008. He notices that while the size of bequests has decreased until the 1950s, this trend is now reversed. Concretely, as a fraction of national income, bequests amounted to 25% in the late 1800s, 5% in 1950 and 15% nowadays. Piketty measures this in two ways, once using wealth data and mortality tables, and once using actual bequests as recorded by fiscal authorities. The second measure is somewhat lower, as expected, but shows the same pattern through time.

The important question is now, what is the new incentive that would make households to leave larger bequests now (and even more in the future, as Piketty suggests)? The old argument of the new growth regime that came with Industrialization is clearly not valid. Piketty proposes that the difference between the rate of return and the growth rate mostly drives the aggregate size of inheritances. But there surely are also important microeconomic aspects, like the wealth distribution, borrowing constraints, the correlation of income between generations and subtleties of the tax code.

What if a cohort lives too long?

Pension funds, life and health insurance companies suffer from a substantial long-term risk: What if their members live longer that expected? While they tend to work with actuarial tables to set premiums and benefits, those tables a re backward-looking, and to price things correctly, the insurers need to make accurate projections about future demographics. But if there is risk, there should be a way to insure oneself against it if a counter-party is willing to take that risk.

David Blake, Tom Boardman and Andrew Cairns make the case that insurance companies should be putting longevity bonds on the market, whose payouts are tied to the demographics of a particular cohort. Imagine an insurance company is selling annuities, but medical developments let some cohorts live unexpectedly longer. By reducing payouts on those cohorts' longevity bonds, the company can finance the shortfall. The question is then who is willing to buy those bonds. Certainly not the cohorts in question. They have annuities already, and those who do not would want a higher payout, not a lower one. Pharmaceutical companies are good candidates, as they face an inverse risk: if a cohort lives shorted than expected, the return from many drugs suffers. But is this enough?