IZA Newsroom

IZA – Institute of Labor Economics

Civility and trust in social media

by

Trust is a fundamental asset for economic development. Social media have been credited with the potential of reinvigorating trust by offering new opportunities for social and political participation. This view has been recently challenged by the rising phenomenon of online incivility, which has made the environment of social networking sites (SNS) increasingly hostile to users.

Online incivility is a manner of harassing behavior that can range from aggressive commenting in threads, incensed discussion and rude critiques, to outrageous claims, hate speech, and more severe forms of harassment such as purposeful embarrassment and physical threats. Pew Research Center (PRC) reports that the majority of Americans have been targeted, or have witnessed others being targeted, with online incivility. The descriptive evidence available so far indicates that social media users perceive incivility as the norm of online interaction.

Given the penetration of social media and the importance of trust in the economic activity, the role of SNS-mediated social interaction is also a relevant topic for economic research. What are the consequences of online incivility and civility on social media users’ trust in others? Does incivility weaken the positive potential of social media?

Experiment in a Facebook setting

To answer these questions, a research team consisting of Angelo Antoci, Laura Bonelli, Fabio Paglieri, Tommaso Reggiani, and Fabio Sabatini conducted a novel experiment in a Facebook setting to study how the effect of social media on trust varies depending on the civility or incivility of online interaction. The study is now available as IZA Discussion Paper No. 11290.

The authors compared the trust and trustworthiness of three samples of participants randomly involved in two kinds of Facebook-mediated interaction. One group was exposed to four, authentic, threads of uncivil discussion. Another group was exposed to the same threads in which uncivil discussions have been replaced with polite interactions by experimenters.

The third group was used as a control condition: participants were exposed to the same thematics used in the other treatments, but in the form of short news excerpts and without any kind of social interaction. To assess trust and trustworthiness, the researchers then made the experiment participants play a trust game.

They found that participants exposed to civil Facebook interaction are significantly more trusting. In contrast, when the use of Facebook is accompanied by the experience of online incivility, no significant changes occur in users’ behavior.

The results of the experiment indicate that when the tone of discussions deviates from the “uncivil status quo” and is accompanied by civil interaction, it significantly raises participants’ trust with respect to both the uncivil treatment and the control condition, whereas it has no effect on trustworthiness. In contrast, if Facebook use is associated with the experience of online incivility in line with the status quo, no significant change in participants’ trust and trustworthiness with respect to the control condition can be observed.

Incivility as the norm of online interaction

This study provides the first experimental evidence of a positive effect of online civility on social trust. While exposure to civil online interaction induced a significant increase in people’s trust, the opposite condition, i.e. online incivility, seemed to be considered “business as usual” and thus did not produce any effect on trust, with respect to the control treatment.

This lack of effect of online incivility is striking, whether it is interpreted as a form of expectation matching (incivility is what people routinely expect online, thus being exposed to it does not change their expectations on others’ behaviors) or as a sort of immunization effect (people are used to such a high level of online incivility that the experimental manipulation failed to elicit a response): in either case, incivility seems to be perceived as the norm of online interaction, rather than the exception.

This is a rather depressing finding, but also one in full accordance with the survey data and the anecdotal evidence available so far. What is much less depressing, and indeed encouraging, is the positive effect on trust of even a brief exposure to online civility: contrary to intuition, according to which a quarrel is much more salient than a polite discussion, a simple lack of aggression in expressing a difference of opinions online acts as a powerful determinant of higher levels of trust towards other people.

Echo chambers and the polarization of public opinion

SNS are facing increasing criticism and scrutiny, since recent analyses of key political events (such as the 2016 US presidential election and the Brexit referendum) have suggested a link between the extreme polarization of public opinion and the relatively small number of platforms that monopolize online discourse – most notably, Facebook and Twitter.

Public discussions on such platforms have also been shown to create and maintain so called “echo chambers”, thus leading to increased polarization and partisanship. Therefore the authors are not entirely surprised to find confirmation, in their results, of a rather bleak outlook on public discourse on SNS, where uncivil debate seems to be considered as normal.

However, the striking result is that even minimal exposure to the opposite trend, i.e. civil online interaction, has a significant effect on social trust. This suggests that what is at stake in moderating online discussion is not simply the prevention of negative phenomena (hate speech, cyberbullying, digital harassment, etc.), but also the achievement of significant social benefits, most notably a measurable increase in social capital that can, in turn, positively affect economic development.

Promoting civil discussion on online platforms

The take-home message for policy makers is rather straightforward: instead of only focusing on fighting against noxious online behavior, we should also (and perhaps mostly) create the preconditions to promote civil discussion on online platforms. Obviously, this goal cannot be effectively pursued via strict regulations, but rather needs to be fostered by carefully designing (or tweaking) the platforms themselves, bringing a wide variety of competencies to bear on such a task: most notably, psychological insight on users’ attitudes and profiles, interaction design principles from ergonomics, nudging strategies and incentives planning from economics.

What drives the gender pay gap?

by

The earnings differential between men and women is a recurring topic in academic research and policy debates. While the gender pay gap is sometimes interpreted as a blatant sign of discrimination that calls for stricter equal pay legislation, the story is far more complicated. For example, women may have a preference for certain sectors, firms and jobs that pay lower wages. Women with children are more likely to interrupt their careers or work part-time. A look at several recent IZA discussion papers shows that these alternative explanations are given different weight, depending on the focus of the analysis and the data used.

Evidence pointing at taste-based discrimination

Using a decade of annual wage and productivity data from New Zealand, an IZA paper by Isabelle Sin (Motu Economic and Public Policy Research), Steven Stillman (Free University of Bozen-Bolzano) and Richard Fabling finds that gender differences in sorting between industries and firms, as well as gender productivity differences, explain just a small fraction of the wage gap. They also reject statistical discrimination as a relevant explanation because the wage gap does not decrease over time even as employers should realize that women are no less productive than men.

The authors conclude that taste-based discrimination is the key driver of the earnings differential. According to the study, this notion is supported by the observation that the wage gap increases under favorable market conditions when employers find it easier to discriminate. The paper thus suggests that stronger enforcement of equal pay regulations could be beneficial for many women in New Zealand, but also in other OECD countries with similar labor markets.

Gender pay gap widens with age

Intra-household decisions in favor of the husband’s career explain why the gender wage gap increases over the life course, as demonstrated in a recent IZA paper by Erling Barth (Institute for Social Research), Sari Pekkala Kerr (Wellesley College) and Claudia Olivetti (Boston College). This is particularly true for college-educated workers, whose earnings profiles tend to be steeper. In terms of higher pay, women tend to benefit less from career moves. The authors use U.S. data to analyze the relative importance of shifts in the sorting of men and women across establishments and differential earnings growth within establishments.

They find that the within component is more important for college-educated workers, explaining three-fourths of their wage gap. The across component – the frequency and quality of job-to-job transitions ­– explains the last quarter for those with college education and the entire gap for those without. The gap is almost non-existent for single women. This may be because married women are often “tied movers”, accompanying their husbands who find a better job in a new city.

Part-time jobs are lower-paid and less stable

Another consequence of the traditional household division of labor is that women are much more prone to part-time work, which typically carries a lower accepted wage rate than full-time work. However, the part-time wage penalty can explain less than 10% (only 3.3% for low-educated workers) of the gender pay gap, according to an IZA paper by Kai Liu (Norwegian School of Economics), meanwhile published in Quantitative Economics. Liu points out that men and women also differ in their job turnover dynamics: women are more likely to quit jobs for non-employment, and job changes for women more often involve changes in hours of work.

Equal Pay legislation for part-time and full-time work would therefore do little to close the gender pay gap, also due to its behavioral impact on labor supply decisions. Instead, the study finds that Equal Protection policies aimed at equalizing the layoff probabilities for part-time and full-time workers would be more effective in reducing the gender wage gap, especially among low-educated individuals.

Willingness to compete affects career choices

So would women earn the same as men if there were no employer discrimination, no household division of labor, and no part-time wage penalty? Probably not, given that gender differences in the willingness to compete still affect career decisions and labor market outcomes, according to an IZA paper by Thomas Buser (University of Amsterdam), Noemi Peter (University of Groningen) and Stefan Wolter (University of Bern).

In an experiment with 1,500 Swiss 8th-graders, the authors find that the gender gap in willingness to compete is essentially zero among the lowest-ability students, but increases steadily with ability and reaches 30–40 percentage points for the highest-ability students. More competitive boys are more likely to choose a math or science-related academic specialization (high-ability boys) or a business-related apprenticeship (medium-ability boys), and are more likely to succeed in securing an apprenticeship position (low-ability boys). These findings relate to persistent gender differences in career outcomes.

The above studies underscore that there are many drivers of the gender pay gap. While there is still disagreement about the major source of it, there is a growing consensus that policies to promote greater female lifetime work effort (e.g. better childcare provision or lower marginal tax rates encouraging female labor force participation) are most effective at achieving more equal pay.

For more information on this topic, explore the IZA World of Labor:

When the unhealthy self kills our New Year’s resolutions

by

“New Year’s Day: Now is the accepted time to make your regular annual good resolutions. Next week you can begin paving hell with them as usual.” — This quote attributed to Mark Twain describes what many people experience as they pledge to stop smoking, work out more often, or eat healthier. But why is it that so many of our New Year’s goals are short-lived?

According to psychologists and economists, lack of self-control causes people to make decisions they will later regret. Food consumption is a leading example of a setting in which self-control problems may play a role. Experimental evidence and the existence of a multi-billion dollar diet industry attest to this. However, there is limited direct evidence on self-control problems from observational consumption data because individuals’ food preferences are heterogeneous and may change over time or in response to changes in the economic environment.

A new IZA Discussion Paper by Laurens Cherchye (University of Leuven), Bram De Rock (University of Leuven), Rachel Griffth (University of Manchester), Martin O’Connell (Institute for Fiscal Studies), Kate Smith (Institute for Fiscal Studies) and Frederic Vermeulen (University of Leuven & IZA) fills this gap with empirical evidence on the existence, size and variation in self-control problems in food choice.

Google search trends reflect intentions to eat healthier

Internet search behavior can tell us a lot about people’s tendency to make (and fail to keep) New Year’s resolutions that would lead more healthy lifestyles: Time trends in Google searches for “diet” and “healthy food” show spikes in January and a steady decline as the year progresses.

Google search trends in the US and UK for “Diet” and “Healthy food”

To find out whether this trend also exists in actual consumption behavior, the researchers exploit longitudinal data on the grocery purchases of a sample of 3,645 single individuals in the UK. The data record all grocery purchases at the transaction level made and brought into the home by these individuals. Using a nutrient profile score, the authors separate the purchased foods and drinks into “healthy” and “unhealthy” baskets.

Healthy January, unhealthy December

The analysis reveals how the share of calories from “healthy foods” varies over time on each day between 2005 and 2012. In line with the Google search pattern, there is a spike in healthiness in January, followed by some decline, plateauing around the middle of the year, then further declining until the end of the year.

Share of calories from healthy foods over 2005-2011

From a theoretical perspective, the variation in diet quality within-person over time can be explained using a dual-self model, in which individual food purchase behavior reflects a compromise between a “healthy” and an “unhealthy” self. These two selves constantly bargain over the food budget and whether to buy fruits, vegetables or whole grains, or to buy unhealthy products such as soda, crisps and confectionery. Increases in the influence of the unhealthy self in decision-making thus indicates failure to exert self-control.

Just over half of calories from healthy foods

The empirical analysis also shows considerable variation in diet quality across individuals: 5% of individuals purchase more than 70% of their calories in healthy foods, while at the other extreme 5% buy less than 35% from healthy foods. On average, people purchase just over half (53%) of calories from healthy foods.

This variation is likely due to both preference heterogeneity and differences in the economic environment people face (such as food prices and budgets). It may also reflect persistent differences in individuals’ propensity to yield to temptation. The findings show that individuals with lower income experience greater variation in their spending on healthy foods, even after controlling for responses to price and budget changes. Younger people, on average, have more variation in their shopping baskets than older people.

Systematic changes also around Easter and birthdays

Easter and own birthdays are two other times of the year which are also associated with systematic changes in purchases of healthy foods. In the run up to these dates, the share of healthy foods purchased tends to decline, gradually over several days in the former case and more starkly a few days beforehand in the latter case. In both cases, the share of calories from unhealthy products recovers towards the pre-event level immediately following the occasion.

Share of calories from healthy foods around significant dates
Image source: pixabay

State-funded mobility assistance helps the unemployed

by

Given that employment prospects vary substantially across regions in many industrialized countries, it seems economically attractive to geographically relocate unemployed job-seekers from depressed to prosperous regions in order to reduce overall unemployment. Governments thus offer financial support to unemployed job-seekers who search for and/or accept jobs in distant regions. Is this money well spent?

In a new discussion paper, Marco Caliendo (University of Potsdam & IZA), Steffen Künn (Maastricht University & IZA) and Robert Mahlstedt (University of Copenhagen & IZA) investigate the impact of Germany’s mobility assistance programs (MAPs) on the job search behavior of unemployed workers and how this affects their subsequent labor market outcomes.

Mobility assistance programs in Germany

Initially introduced in 1998, MAPs in Germany consist of six separate programs ranging from reimbursement for distant job interviews to relocation assistance. For example, travel cost assistance reimburses expenses for distant job interviews up to an amount of €300. The daily commute to a new job is financially supported with 20 cents per kilometer for the first six months. The relocation assistance program for those who permanently move to a new location provides full coverage of transportation costs up to €4,500.

Using the IZA Evaluation Dataset Survey, which comprises information on more than 17,000 individuals who entered unemployment between June 2007 and May 2008 in Germany, the authors were able to empirically analyze the impact of MAPs on job search behavior and labor market outcomes of these individuals.

Larger search radius

Regional differences in the promotion of MAPs, in combination with the detailed survey data, allow the researchers to analyze how the existence of these programs affect the job search behavior. The results show that reducing the costs of distant job search induces job-seekers to extend their search radius.

The increase in distant job search effort due to MAPs leads to a 16 percentage points higher probability of being regularly employed 12 months after entry into unemployment. Distant job seekers also realize significantly higher hourly earnings in the following job compared to local job seekers (15 percentage points), but they work fewer hours per week, resulting in an overall zero effect on monthly earnings.

Better promotion and more funding needed

It is important to note that the positive employment and earnings effects are generated by those job-seekers who actually become geographically mobile (either by commuting or relocating) and accept jobs in distant regions. Therefore, the study concludes that the German system of MAPs is indeed effective in bringing job seekers back to work by triggering them to increase their search radius, and providing financial support to accept geographically distant jobs. According to the authors, this is a “promising finding” in the light of relatively lost costs per participant.

This means for policy makers that better information about the availability of the programs must be accompanied by an increased budget for MAPs. The authors also suggest that the introduction or expansion of MAPs may be worth considering to stimulate geographically mobility not only within, but also between European countries to reduce disparities in unemployment rates.

See also a related previous paper by the same authors (IZA DP No. 9183), which was published in the Journal of Public Economics and covered in the German IZA Newsroom:

Image source: pixabay

Do boys benefit from male teachers in elementary school?

by

boyThe scarcity of male teachers in elementary school has led to a debate on whether young boys require more male role models in school, and whether boys might be discriminated against by female teachers. A recent IZA Discussion Paper by Patrick A. Puhani (Leibniz University Hannover) suggests that such fears are unwarranted.

Using administrative data on the population of students and teachers in the German state of Hesse, Puhani estimates teacher gender effects on elementary school outcomes. The evaluation of within-school variation, however, which controls for school fixed effects, identifies virtually no effects of teacher gender at the end of elementary school (grade 4, age 10) on either the teachers’ recommendations for middle school type choice or the actual school type choice in Germany’s early tracking system.

The one exception is that boys might benefit slightly in terms of a higher school type recommendation when taught by a male teacher, although there is no such effect on actual school type choice, probably because parents have the final word on this matter. Nor do the teacher fixed effects models reveal any effects of being taught by a teacher of the same gender on either outcome variable.

Puhani thus concludes that his findings “should allay the concerns expressed in the global press that the increasing feminization of elementary school education might lead to systematic discrimination of boys.”

Image source: pixabay

Economic costs of global warming

by

Climate change is considered one of the major challenges of the 21st century. While politicians and scientists at the COP23 climate change conference in Bonn are currently debating the political implications, economists have looked into the various economic consequences of global warming. Three papers co-authored by IZA’s “Environment and Labor Markets” Program Coordinator Olivier Deschenes investigate the impact of rising temperatures on productivity, birth rates and mortality.

The authors of the most recent paper, Peng Zhang (Hong Kong Polytechnic University), Olivier Deschenes (UC Santa Barbara & IZA), Kyle C. Meng (UC Santa Barbara), and Junjie Zhang (Duke Kunshan University), analyze detailed production data from a half million Chinese manufacturing plants over 1998-2007 to estimate the effects of temperature on firm-level total factor productivity (TFP), factor inputs, and output.

Chinese manufacturing output could fall by 12% per year

The findings show that both labor- and capital-intensive firms are affected by high temperatures. According to the study, one day with temperature above 90 degrees Fahrenheit (32° C) reduces TFP by 0.56% and manufacturing output by 0.45%, or by $8,160 in 2007 dollars, for the average firm. Since the temperature effects on labor and capital inputs are not as pronounced, TFP losses in response to high temperatures appear to be the primary channel through which temperature alters manufacturing output.

Based on historical and projected future climate data, the authors calculate that until the 2040-2059 period, average temperatures in China will increase by 3.6°F (2.0°C). As the map shows, more extremely hot days are to be expected especially in eastern and southern China.

Change in days per year with temperatures above 90°F between 1998-2007 and 2040-2059 periods

For the Chinese economy, this means that climate change could reduce manufacturing output by as much as 12% annually by mid-21st century or by $39.5 billion in 2007 dollars if no adaptations are undertaken. If China’s manufacturing output share remains fixed at 32% of national GDP, those predicted climate-driven losses in manufacturing alone would reduce Chinese GDP by 3.8% annually by mid-century, according to the study.

Global warming also affects birth rates and mortality

The findings of IZA Discussion Paper No. 9480 show that demography is also affected by global warming. The authors, Alan Barreca (Tulane University & IZA), Olivier Deschenes, and Melanie Guldi (University of Central Florida), conclude that increased temperatures due to climate change may reduce population growth rates over the course of the century. They estimate the effects of temperature shocks on birth rates in the United States between 1931 and 2010 and find that additional days above 80°F (27° C) cause a large decline in birth rates approximately 8 to 10 months later.

For more information on this topic, see the IZA World of Labor article by Alan Barreca: Does hot weather affect human fertility?

A third paper by Alan Barreca, Karen Clay (Carnegie Mellon University), Olivier Deschenes, Michael Greenstone (University of Chicago), and Joseph S. Shapiro (Yale University) evaluates the relationship between rising temperatures and life expectancy. In the economics literature, hot temperatures have been associated with excess mortality due to cardiovascular, respiratory, and cerebrovascular diseases.

The paper shows that the mortality impact of days with a mean temperature above 80° F has declined by about 70% after the 1960s. The results show that the drop in mortality can almost entirely be explained by the adoption of residential air conditioning in the US. In contrast, electrification and access to health care were not significantly related to changes in the temperature-mortality relationship.

The authors conclude that existing technologies (such as residential air-conditioning) offer tremendous opportunities to mitigate the impact of climate change. At the same time, greater use of energy-intensive adaptations may speed up the rate of climate change because electricity production worldwide continues to be primarily driven by the combustion of fossil fuels. “The continued expansion of the renewable energy sector and the development of more efficient adaptation technologies should also be key components of the world’s climate change strategy,” says Olivier Deschenes.

Image source: pixabay

Low-skilled immigration causes human capital polarization in Italy

by

While there is a vast literature considering the labor market effects of immigration, less has been done to investigate how immigration affects the educational choices of young natives, especially in Europe.

A new IZA discussion paper by University of Padova researchers Giorgio Brunello, Elisabetta Lodigiani and Lorenzo Rocco looks at the impact of low-skilled immigration on the educational choices of young natives in Italy during the period 2006–2016. The case of Italy is interesting because the majority of immigrants originate from developing countries and are low-skilled. Also, immigration to Italy is a recent and rapidly growing phenomenon. During the period of interest, the average share of immigrants almost doubled, from 3.7 to 7 percent of the population.

Decreasing marginal benefits for intermediate education

The authors argue that one possible outcome of higher immigration is human capital polarization – or the contemporaneous increase in the share of natives with lower (less than high school and not in education) and higher education (enrolled in college or with a college degree). They document that polarization is not only a theoretical possibility but also an empirical fact. The evidence for Italy – stronger for males than for females – indicates that the recent inflow of low-skilled immigrants has increased the share of both lower educated and higher educated natives.

The study also shows that native males who choose not to invest in further education because of immigration are more likely to work in manual jobs and in the service sector. Native females are instead more likely to be inactive. In line with previous literature, natives who invest more in education because of immigration are less likely to choose STEM fields and more likely to enroll in “communication-intensive” fields.

Reduction of the middle class

The findings indicate that immigration is another source of polarization, as much as globalization and technological progress. The economic literature so far has emphasized the latter but has almost overlooked the former. In the setup of the paper, the polarization of educational choices is privately optimal, but has unpalatable aggregate consequences, because it reduces social cohesion by reducing the size of the middle class. Importantly, the less privileged class is unlikely to consist only of immigrants. By pushing many natives out of school too early, low-skilled immigration is contributing to the expansion of a native underclass.

These results raise questions about what sort of immigration policy a country should select. On the one hand, attracting cheap unskilled labor from abroad can help support an industrial structure that relies more on the price of labor than on technological innovation. On the other hand, by delaying innovation and by reducing the human capital investments of many natives, this policy can have negative consequences on long-term productivity and international competitiveness and contribute to economic decline.

Fertility and the digital divide: More flexibility, more children?

by

The rapid diffusion of the Internet, and in particular of high-speed, broadband Internet, has characterized the life-changing digital revolution that began around the turn of the twenty-first century. The profound social and economic implications of the spread of the Internet have been highlighted since its inception by social science scholars. More recently, researchers have also started to focus on the effect that the Internet has on family life. Scholars interested in family and fertility have long been focused on the importance of technological change.

Shifts in contraceptive technology, alongside the development of household appliances and medical advances, have been pointed to as some of the preconditions for the massive reproductive changes and the parallel large increases in women’s education and labor force participation that took place in high-income societies during the last period of the twentieth century. The discussion on low fertility, however, has not as of yet focused on the role of broader technological change and how it could shape the future of demographics, with the digital revolution epitomizing such technological change.

DSL technology boosts fertility of highly educated women

In a new IZA discussion paper, Francesco C. Billari (Bocconi University), Osea Giuntella (University of Pittsburgh and IZA) and Luca Stella (Bocconi University and IZA) analyze the impact of the diffusion of high-speed Internet on fertility choices in a low-fertility setting, Germany.

The authors exploit an instrumental variable approach devised by Falck et al. (2014) relying on unique historical and technological peculiarities of the public telephone infrastructure in Germany which affected the diffusion of DSL technology throughout the country. Using data from the German Socio-Economic Panel (2008-2012), they show that DSL access increases fertility of 25-45 year-old women and that these results are largely driven by highly educated women. The rise in fertility mostly reflects an increase in the probability of having more than one child.

Internet access may help bridge competing work-life goals

Moreover, the researchers investigate the potential mechanisms underlying the relationship between Internet and fertility. Their results suggest that broadband may increase fertility by increasing the opportunities of working from home and/or working part-time. These effects may relax time constraints, especially among more educated women, thereby favoring the work-family balance. They conclude that broadband might introduce a “fertility digital divide,” allowing highly educated individuals to realize their fertility goals, while not improving the chances of low-educated individuals, who tend to be employed in less flexible occupations.

Image source: pixabay

Teenage daughters as a risk factor for divorce

by

Sullen exchanges, inexplicable silences and broken curfews can be part of life for parents of teenagers, but could this period be also a stress-test for parents’ marriages? A new IZA Discussion Paper finds that parents of teenage daughters are more likely to separate than parents of teenage sons. We wanted to know more from the authors, IZA fellows Jan Kabatek and David Ribar of the Melbourne Institute.

Jan Kabatek

What is the main finding of your paper?

Our research studied more than two million marriages in the Netherlands over twenty years and showed that divorce risks faced by Dutch couples are dependent on age and gender of their children. The risks increase with children’s ages up to the point when children reach adulthood, and parents of teenage daughters are at greater risk still.

Has this topic been researched before?

David Ribar

The associations between marital strains and children’s gender were first identified by American sociologists in the 1980s. Economic research followed later, and several studies in the United States have indeed found that parents with first-born girls are slightly more likely to divorce than parents with first-born boys. However, until now, there was no evidence from other developed countries which would confirm that daughters strained marriages, and the heterogeneity of the gender effect remained largely unexplored.

Your study uses registry data from the Netherlands. Why Dutch data?

Compared to the data used in most previous studies, the Dutch administrative records are advantageous in several ways. They allow us to analyze a very large pool of marriages, the sample selection issues are minimal, and the records themselves are exceptionally comprehensive. We can look at exact dates of weddings, births, and divorces and delve deeper than studies which relied on self-reports and people’s recollections. More importantly, the data also allow us to link all parents to their children and examine just how long after their birth the couples separated.

What was the most striking result?

We found that the gender effect does exist among the Dutch parents, however it is strictly confined to the teenage years. Up until the age of 12, the gender of the child has no influence on the divorce risks faced by the parents. It is only between the ages 13 and 18 that parents of first-born girls divorce more frequently than parents of first-born boys. This finding contrasts prior evidence from US censuses, which documented significant differences of divorce risks for American families with children aged 0-12.

How does the teenage effect translate into numbers?

Conditional on staying married throughout the first twelve years of the first-born’s life, the odds of divorce are 10.7% for parents of teenage boys, and 11.3% for parents of teenage girls. In relative terms, this means that parents with teenage daughters face 5% higher risks of divorce than parents with teenage sons. The effect peaks at the age of 15, when the risk faced by parents with daughters is almost 10% higher than the risk faced by parents with sons. In the following years, the differences narrow again, and they disappear once the child turns 19. A similar pattern is also found among second-born and subsequent children.

Does this mean that Dutch parents have a preference for sons?

We don’t think so. The null finding for families with young children goes against the standard son-preference hypothesis, which implies that the mere presence of male children would make the marriage stronger. Furthermore, son preference would also influence fertility levels, rendering them higher for families with first-born girls. However, similar to recent US evidence reported in another recent IZA discussion paper, we find the exact opposite: families with first-born girls have slightly fewer children than families with first-born boys.

So what are the reasons why daughters might raise divorce risks?

We do not find evidence supporting several other well-established arguments, such as the theory which assumes that boys are more vulnerable and their need of male role models makes fathers more committed to the marriage. The same is the case for a sex-selection theory which postulates that mothers whose marriages are more stressful may be more likely to give birth to a baby girl.

Instead, our findings suggest that the higher divorce rates are explained by strains in the relationships between some parents and their teenage daughters, possibly stemming from differences in attitudes to gender roles. This explanation is backed by analysis of a large survey of Dutch households, which asked families about their relationships and opinions regarding marriage, gender and parenting.

What did the families say?

Parents of teenage daughters disagreed more about the way they should raise their children, and expressed more positive attitudes towards divorce. They were also less satisfied with the quality of their family relationships. Teenage daughters, in turn, reported worse relationships with their fathers, though not with their mothers.

Dads just don’t connect with their daughters?

Such a statement would be taking the empirics too far. Our findings do, however, suggest that the relationship of the father and his daughter is an important piece of the puzzle.

In one exercise, we split the administrative sample into two groups, depending on whether the father did or did not grow up with a sister. Our hypothesis was that the fathers who had more experience relating to teenage girls via their sisters would experience fewer relationship strains with their own teenage daughters. This could occur because fathers with sisters may hold more egalitarian attitudes towards gender roles, or because they have a better understanding of teenage girls and their family interactions.

In line with this reasoning, we found that the fathers who grew up with sisters did not face any increase in divorce risks from teenage daughters. The gender effect only appeared among fathers who grew up without sisters.

Do other family characteristics play a role?

We looked at other characteristics that could indicate differences between the gender-role attitudes held by the parents and their daughters, such as the ages or immigration background of the couple. Here, we also found that the parents who are likely to hold more traditional attitudes towards gender-roles experienced higher increases of divorce odds from teenage daughters.

Are the increased odds of divorce from teenage daughters unique to Dutch married couples?

It doesn’t seem so. We find the same associations for Dutch couples in de facto relationships, and our analysis of the Current Population Survey confirmed their existence also among married couples in the US. Importantly, we show that while we cannot reject the existence of a small gender effect for US families with children aged 0-12, we can demonstrate that such an effect is clearly dominated by the disparity that emerges in the teenage years. Compared to the previous estimates of the gender effect for US families with young children, the teenage effect derived from the CPS data is more than ten times larger. It is also substantially larger than the teenage effect found in the Dutch data.

So should parents of girls be worried they might be destined for divorce?

Not really. Despite their relative significance during the teenage years, the differences in the divorce risks faced by families with boys and girls remain modest over the child’s lifetime. By the time their first-born children reached age 25, 311 of every 1,000 Dutch couples with daughters had divorced, which can be compared to 307 of every 1,000 with sons—a difference of only 4 divorces per 1,000 couples.

Furthermore, the finding of a null effect among fathers who grew up with sisters shows that the association between the children’s gender and divorce risk is not universal. That being said, our study does point to serious strains between some parents and their teenage daughters, and help us understand the factors contributing to family break-down.

What can parents do to reduce these risks?

Our results suggest that parents of teenage daughters would do well to adopt more egalitarian attitudes towards gender roles and a greater understanding of how conflicts could come up. Struggles with teenagers will still happen, but better preparation and knowledge of the wants and needs of their teenage daughters could reduce the strain between partners.

Providing our children with role models who have modern attitudes toward gender roles, and promoting open communication within the family unit may contribute to a lowering of the rate of divorce or separation.

Thank you very much!

Teilkrankschreibungen können betriebliche Fehlzeiten und Gesundheitskosten reduzieren

by

Bei Krankschreibungen gilt in Deutschland bislang die Alles-oder-Nichts-Regelung: Arbeitnehmer sind entweder arbeitsunfähig oder müssen die volle Arbeitsleistung erbringen. In der Realität sind krankgeschriebene Beschäftigte aber häufig willens und in der Lage, schrittweise und mit verringerter Stundenzahl in den Job zurückzukehren. Immer wieder wird daher das Modell der Teilkrankschreibungen diskutiert, das in Deutschland selten und nur auf freiwilliger Basis Anwendung findet.

Wie sich eine gesetzliche Verpflichtung zu Teilkrankschreibungen auf betriebliche Fehlzeiten und das Krankengeld auswirkt, untersucht ein aktuelle Studie von IZA-Fellow Øystein Hernæs (Institute for Social Research, Oslo) auf Basis einer in Norwegen durchgeführten Reform. Dort ist es seit 2004 gesetzlich vorgeschrieben, dass Krankgeschriebene so früh wie möglich, in der Regel nach spätestens acht Wochen, wieder teilweise in den Job zurückkehren.

Da die Reform jedoch in der Praxis nicht konsequent durchgesetzt wurde, entschied sich die Regierung der Region Hedmark, mit einer Reihe von Maßnahmen nachzuhelfen: Sachbearbeiter wurden geschult und Krankgeschriebene regelmäßig ermutigt, ihren Arbeitgeber wegen einer Rückkehr in Teilzeit zu kontaktieren. Zudem wurden Ärzte, Unternehmen und die Öffentlichkeit mit einer Informationskampagne aufgeklärt. Da diese Aktivierungsmaßnahmen regional begrenzt waren, konnte Hernæs die Effekte isoliert betrachten.

Schnellere vollständige Rückkehr in den Job

Die Ergebnisse zeigen deutlich die positiven Effekte des Programms: Betriebliche Fehlzeiten wurden um 12 Prozent verringert und das ausgezahlte Krankengeld reduziert. Die positiven Effekte stellten sich unabhängig von Geschlecht, Alter und Branche ein – nur nur im Baugewerbe fielen sie geringer aus. Am stärksten waren die positiven Arbeitszeiteffekte bei Arbeitnehmern mit Muskel-Skelett-Erkrankungen. Die positiven Effekte blieben jedoch nicht nur auf die in Teilzeit geleisteten Stunden beschränkt: Teilkrankgeschriebene Arbeitnehmer kehrten schneller wieder vollständig in den Job zurück.

Die Studienergebnisse zeigen somit die positiven Potenziale von Teilkrankschreibungen. In Deutschland besteht zwar mit dem sogenannten Hamburger Modell bereits jetzt die Möglichkeit, nach sechs Wochen Krankheit auf freiwilliger Basis wieder einige Stunden zur Arbeit zu gehen und in dieser Zeit weiterhin Krankengeld zu erhalten. Vorschläge zur Einführung einer gesetzlichen Grundlage zur Teilarbeitsunfähigkeit wurden jedoch bislang nicht weiter verfolgt.