Are you willing to take a risk? : a panel study of cohort change in risk attitudes in West Germany 1984-2016

(1)

1

Are you willing to take a risk? A panel study of cohort change in

risk attitudes in West Germany 1984 - 2016

MSc. Sociology (general track)

University of Amsterdam

Ingrid Glas

Student ID: 10793682

8

th

July 2019

Ingridglas@gmail.com

Supervisor: Thomas Leopold

Second Reader: Thijs Bol

(2)

2

Attitudes towards risks are an extensively studied topic as the willingness to take risks is a crucial determining factor of individual choices in several domains. However, little attention has been paid to generational change in the willingness to take risks. This study aims to contribute to the existing literature by studying generational change in risk attitudes under the West German population. By making use of the third version of the German Socio-Economic Survey Panel study it is possible to disentangle the effect of age and cohort on risk attitudes. Because prior research indicates there is a gender gap in risk attitudes, gender is included in the model and is interacted with the separate cohorts to study if there are any gender differences between different cohorts. The sample population in this study consists of 19 704 individuals ranging from the ages of 17-87 born in or after 1920. A generalized ordinary least square random effect model is used to study the separate effects of age and cohort on the willingness to take risks. The results indicate that age has, as expected, little effect on the general willingness to take risks. Secondly, the willingness to take risks is increasing with every successive cohort, with a large increase between the oldest and the youngest cohort of 2.271 scale points (on an eleven-point scale) with an effect size of 0.98. Thirdly, the cohorts that experienced the war during an impressionable age (the cohort of 1920 and 1930) are least willing to take risks compared to the other cohorts. Fourthly, there is a relatively big leap in the general willingness to take risks between the two older cohorts and the cohort of 1940. And finally, women are on average more risk averse than men, these differences do not decline in time.

(4)

4 1. Introduction

“The biggest risk is not taking any risks.” These are the words of the owner of one of the most influential social platforms of our time: Mark Zuckerberg (Tobak, 2011). And his statement might not be so far from the truth. Risk taking is a very important aspect of life. Without ever taking any risk you will probably not leave the safety of your current job and apply for that exciting job you always wanted. Or leave everything behind to book that journey that you are longing to make. But on the other hand, taking too much risks, like excessive drinking, smoking or gambling, can make your life very difficult. Your willingness to take risks consequently plays a key role in political, demographic and economic decision-making processes (Dohmen, Falk, Golsteyn, Huffman & Sunde, 2017). The willingness to take risks impacts political choices, where risk tolerant individuals tend to vote more progressively (Kam & Simas, 2012). It influences economic and investment choices where risk tolerance is associated with investments in risky assets. Furthermore, a willingness to take risks plays an important role in marriage and fertility decisions (Schmidt, 2008). Consequently, understanding changing attitudes regarding risk taking is intimately linked to the goal of understanding and predicting behaviour in several domains.

Most research about risk taking is however conducted by economy or psychology scholars, who are interested in financial risk-taking behaviour or individual change in the willingness to take risks throughout the life course. Consequently, little attention has been paid to societal change in the willingness to take risks. Where standard economic models assume that humans are rational beings who weigh all the historical information when they form a believe about risky outcomes. Psychology scholars, on the other hand, argue that personal experience exert a greater influence on personal believes and decisions than factual information. More recently, there is a growing body of literature which indicates that the willingness to take risks of a group of individuals, for example those born in a certain time period, may also change under influence of shared external macro-events. For example, depression babies, individuals that were born during the Great Depression in 1930, were brought up in a different historical context than people born during the flourishing 60’s (Malmendier & Nagel, 2011). Consequently, the birth cohorts that entered the labour market during these years of economic depression became less willing to take risks than the generations who came after them.

(5)

5

willingness to take risks, there are, to my knowledge, no scholars who study generational change in the willingness to take risks with panel data. Most scholars study the relation between socio-demographic variables, like level of education or marital status, and the willingness to take risks, or study the willingness to take risks throughout the life course (Dohmen, Falk, Golsteyn, Huffman & Sunde, 2017). This gap in the literature requires a different perspective that contributes to the already existing literature.

Therefore, the present study adopts a different approach and uses panel data from the German Socio-Economic Panel study (SOEP) to study generational change in the willingness to take risks among the West German population. For clarification, the terms ‘cohort’ and ‘generation’ shall be used interchangeably in this paper to represent individuals born at the same, or approximately the same time (Elder & George, 2016). Moreover, the terms ‘willingness to take risks’ and ‘risk attitudes’ shall be used interchangeably to represent how willing individuals are to take risks, where risk tolerant attitudes lead to more willingness to take risks and risk averse attitudes lead to less willingness to take risks. This study addresses the question if West German society is becoming more, or less willing to take risks and what kind of shared societal experiences could have caused this change in attitudes towards risk taking? Furthermore, as prior research indicates that the willingness to take risks differs between men and women, a distinction will be made based on the sex of the respondents. The German Socio-Economic Panel study is especially useful to investigate this question because it is a large-scale panel dataset which has been conducted annually since 1984. The question about general willingness to take risks has been included since 2004. The questionnaire runs up to 2016 which means it covers a time period of 13 years in which individuals are questioned in overlapping ages.

The paper is organized as follows. First, a chapter about social relevance will elaborate more on why studying generational change in risk attitudes is important. Next a conceptualization of the notion of risk will be provided to offer the reader some guidance. The chapters that follow will elaborate more on societal changes in West Germany and through which mechanism these changes potentially cause a change in the willingness to take risks for different birth cohorts. These theories are then used to formulate hypothesis. Subsequently, the German Socio-Economic Panel study is introduced, and the model that I will use to investigate the main research question will be discussed in further detail. Finally, the results will be presented and discussed along the lines of the literature to formulate an answer to the research question.

(6)

6 1.2. Social relevance

The importance of attitudes towards risk taking in the decision-making process is not to be underestimated. As mentioned above, the willingness to take risks is an important determining factor of individuals choices in several domains. In the political domain, voting is sometimes compared to gambling. Voters can never be completely sure that the candidates, after their election, will bring their standpoints into practice. Therefore, there is always an element of uncertainty and risk. Individuals who are more risk tolerant appreciate a certain amount of sensation and excitement and are therefore more likely to vote for change. Consequently, risk takers vote significantly more often for house challengers and are more drawn to candidates who offer less certainty. While risk averse individuals generally search for more stability and certainty and are therefore less likely to vote for the unknown. Although it is important to note that individuals in general, whether they are risk averse or risk taking, are less likely to vote for political newcomers (Kam & Simas, 2012). Moreover, individuals who are risk tolerant tend to have more progressive viewpoints regarding free trade or migration (Erlich & Maestas, 2010) and tend to support military interventions more often (Eckles & Schaffner, 2011). Consequently, whether West German society becomes more, or less willing to take risks has an important impact on how the political landscape would look like. If West German society is becoming more risk tolerant, this could make the political landscape more dynamic with more sudden changes. Individuals could become more willing to vote for political newcomers. This might result in many new political parties who get elected for a shorter time. Furthermore, an increase in risk tolerant attitudes towards risk taking could mean that liberal changes or daring policies get more support than before. In the other scenario, when German society is becoming more risk averse, it could mean that individuals vote for political parties with more conservative standpoints, potentially increasing resistance to reforms that might be necessary.

The same applies for economic choices. Saving and investment decisions are shaped by a willingness to take risks, and risk tolerant individuals tend to spend more and save less (Bommier, Chassagnon & Le Grand, 2012). Occupational sorting is also influenced by a willingness to take risks. Some jobs have higher unemployment risks or earnings risks, and risk tolerant individuals are usually more willing to take these risks and are also more likely to become self-employed (Bonin, Dohmen, Falk, Huffman & Sunde, 2007; Caliendo, Fossen, & Kritikos, 2007). The balance between saving and spending is very important for a healthy economy, because our economy is

(7)

7

based on a certain amount of spending. Moreover, if occupational sorting is influenced by a willingness to take risks, this potentially affects inequalities between groups.

Finally, the willingness to take risks has a significant effect on timing of marriage and fertility decisions (Schmidt, 2008). Women who are risk tolerant are more likely to delay marriage. The underlying idea is that searching for the perfect partner is a form of risk taking, eventually, if you wait too long it can result in not finding a suitable partner or discovering you gave up on the one you like the most. From another angle, marriage is seen as self-insurance against income-risks. Although there are also theories that predict the opposite, Schmidt (2008) finds, in line with the above described theories that women who are more risk averse, marry earlier than women who are more risk tolerant. Risk tolerance is also correlated with earlier births at a young age. It is argued that this effect is caused by these women being less likely to contracept effectively (Ibid). The picture differs when college-educated older women are considered. Then, risk tolerant women are likely to delay childbearing and therefore are less likely to become mothers. If West German society is getting more willing to take risks this could mean in increase in singles. If marriage delay is used to find the perfect partner, then it is likely that on the quest for the right partner, individuals have many short-term relationships before marriage. This influences also other, on first sight not related, area’s like housing and the labour market. Moreover, it would imply an increase in (young) single mothers.

(8)

8 2. Theoretical Framework & Background

2.1 Conceptualizing Risk

As the willingness to take risks is the primary focus of this study, it is important to define what exactly is understood with the notion of ‘risk’. In the Cambridge dictionary, risk is defined as “The possibility of something bad happening” (Cambridge Dictionary, 2019). But this gives rise to the question; what is “something bad” and is this the same for everyone and across every space and time?

The term ‘risk’ is first found around the 12th/13th centuries, where it was mostly applied to the emerging insurance-business to conceptualize a change in structure. In the context of a seemingly increasing controllable “Nature”, instead of attributing possible losses to external factors, like God, Nature or Destiny, individuals were increasingly held responsible. The concept ‘risk’ reflected this changing perception of the world and emphasised the computability of the world around you (Voss, 2004, p.2). In the present study, this risk calculation is grounded in perception. Risks are not seen as objective facts, instead it is perceived as a socially constructed concept. The notion of risk is negotiated in the social realm where it is shaped by the actual social context, social experiences and imagined futures (Zinn, 2017, p.4). This means that risks are not always rational. Some risks are shaped by the actual social context, like structural labour market insecurities that heightens the chance of job loss. Others are shaped by imagined futures, which is partly a fictional expectation. Climate change is an example of a risk that does not pose immediate danger, and for some people it will never really threaten their lifestyles, but it does threaten our imagined future and thus shapes risk perceptions. The same can be said about terrorist threats; the chance of being the victim of a terrorist attack is very small, but the renewed violence terrorism brings with it, alters the way we imagine our society. On an individual level, our imagined future is also shaped by role expectations. Individuals construct their expectations and behaviour according to societal expectations about their social position. A transition into parenthood or marriage, for example, changes someone’s social position which alters perceptions of gains and losses and therefore causes a change in attitudes towards risks (Chaulk, Johnson & Bulcroft, 2003). Although the above examples seem to imply that risk is exclusively a negative concept, this is not necessarily the case. Risk taking can be a thrilling and exciting activity. And taking certain amounts of risks is often perceived as a good quality.

(9)

9

To get a better grip of what the concept entails, it is useful to relate it to another concept that is closely related; uncertainty. Strictly speaking, when the outcome of one’s actions is absolutely certain, there is no room for risk. So, every conception of risk requires some uncertainty. But risk and uncertainty are not the same. In social theories about risk, risk is often seen as a rational strategy to transform unmanageable contingencies into manageable complexities (Zinn, 2006, p.6). In this view, the world is inherently contingent, and by making use of past experiences and the social context as a source of information, an attempt is made to search for patterns to manage these contingencies. Thus, risk taxation, in this view, is a strategy to calculate and manage uncertainties. Furthermore, when individuals calculate potential risks, relative to prior experiences, the social context and imagined futures, perceived losses weigh twice as heavy as perceived gains and securities more heavily than chances (Chaulk, Johnson & Bulcroft, 2003).

In the present study, risk calculation is perceived as a strategy to manage uncertainties. Therefore, risks are not objective facts, instead risk taxation is an important decision-making process based upon an individual’s frame of reference (Wright, 2017). In creating this frame of reference, the social context is of vital importance. Therefore, the shared social context of West German birth cohorts is of primary importance in this study. A shared social context also creates, to a certain extent, shared experiences and individuals who grow up in the same birth cohort are raised with somewhat similar norms, values and role expectations. As perceived losses weigh higher than perceived gains and securities are more important than opportunities, negative circumstances, should have a greater impact on risk attitudes than positive circumstances. Although risk taxation is grounded in perception, attitudes towards risks also get more or less crystalized at a certain age. I will elaborate more on how attitudes get shaped in a later chapter, for now it is important to state that a stable component of the attitude towards risks is shaped during young adulthood.

2.2 Studying societal change

A useful method to study generational change is by analysing cohorts. A cohort is a group of individuals who experience an event at the same time. Cohorts can be defined according to all kinds of significant events, like going to college or entering the labour market (Elder & George, 2016). In this paper, a cohort specifically refers to birth cohort. An individual’s birth year locates

(10)

10

them in a historical context and time that they share with individuals born in the same year. Although there are individual differences, like, race-ethnicity, social class or gender, individuals who share their birthyear grow up in a specific social context and will go through important life phases somewhat around the same time. Norman Ryder was one of the pioneers of studying societal change through cohorts and proposed that cohorts are the vehicles that bring about social change. In Ryder’s words “Cohorts do not cause change; they permit it. If change does occur, it differentiates cohorts of individuals from one another, and comparison of their careers or life trajectories then becomes a way to study change” (1965, p.845 in Elder & George, 2016). For Ryder it is not only a possibility that cohorts trigger social change, it is virtually a certainty that they do. Thus, a birth cohort is more than just a sum of individuals born in the same year; it reflects unique circumstances in one point in history. Cohorts that experienced major social disruptions during young adulthood or adolescence, such as a war or major economic changes, generate more social change than cohorts with milder circumstances when coming of age. Furthermore, cohort size can affect the nature and volume of social change. When a cohort consists of many individuals, for example the baby-boom generation, they have the numerical power to generate more change and imprint this more firmly on the rest of society in comparison to smaller cohorts (Ibid.). Therefore, it is important to briefly outline the West German social context and identify if any social disruptions could have had a long-lasting impact on the risk attitudes of different birth cohorts. For this research, birth cohorts of 10-year intervals are used to study how West German risk attitudes have changed across time, where the first cohort is born in 1920 and the last in 1990. The next chapter will elaborate more on the social history of Germany.

2.3 Changes in German society

To outline some relevant changes, and its consequences, in German society, we will go back to the 1930s, where there was a world-wide economic crisis called the Great Depression. Germany was, together with the United States, the hardest hit by the crisis and this economic crisis is still a crisis without parallel in the history of industrial capitalism. Besides the falling production rate there was a sharp increase in unemployment. At the peak of the crisis, in 1932, there were officially 6 million unemployed in Germany. However, many unemployed did not appear in the official records because many people who were unemployed no longer received support after years of

(11)

11

unemployment and were no longer included in statistical records (Petzina, 1969). Petzina (1969, p.3) estimated that in 1932 one in every three of the working population was without a job and those with jobs were under constant threat of dismissal. Although none of the respondents in this sample was part of the working population during the Great Depression, their parents were, and the Great Depression had an impact on family structures. During the Great Depression fertility rates were low and traditional gender roles shifted. In the 1930s men were expected to be the breadwinners of their families but due to the high unemployment rate, many were unable to fulfil this task. Many women increasingly came to work outside of the home in low-skilled and often low-paid jobs to support their families while the men stayed at home (Bryson, 2004).

This poor economic position of Germany led to social unrest which in turn helped to create political climate where extreme political parties were emerging. This resulted in a growth in power from the NSDAP, the political party of Adolf Hitler who eventually led Germany into war in 1939. Understandably, World War II had a huge impact on the German society. During World War II there were important differences between the role of men and women. Hitler heavily believed in traditional gender roles where women stayed at home to care for the children and their home while men were working to financially support the family. Moreover, family life in general was encouraged during the war. This resulted in a strong institutional discouragement for women to enter the labour market. Female employment during the war dropped while marriage rates and birth rates increased (BBC, 2019). Although during the first years of the war there were no severe shortages of food or clothes, which was partly due to the exploitation of occupied countries, this situation changed following the end of the war when the shortages became severe (Ibid.). Shortly after the war, Germany was in chaos. Its economy was almost totally destroyed, causing chronic unemployment and poverty (Davis, 2015). The food shortage still lingered on, there was no central government to implement new laws and millions of people were homeless or attempting to return to homes that no longer existed (Knowles, 2013). All these processes had a deep impact on everyday life in Germany. Many certainties, such as finding a permanent job, having enough food or renting a house, became uncertain, which can result in mental stress. Another result of the war which can easily be overlooked but had a major impact on the life course of the cohort of 1920, was a highly distorted sex-ratio. Due to upheavals during and after the war, many soldiers died, there were high rate of imprisonment, mass emigration and resettlements. For the German cohort that was born in 1920 there were only 73 men to 100 women at age 36. This, at least partly, resulted

(12)

12

in high levels of childlessness among the German population (Kreyenfeld & Konietzka, 2017). Furthermore, an unbalanced sex-ration also led to lower marriage rates and an increased divorce rate (Chipman & Morrison, 2013). The war also had a long-lasting impact on the belief systems of its citizens. A national sentiment of guilt and shame formed German society long after the war was over and a certain scepticism about military interventions can still be observed in the younger generation (Davis, 2015).

From 1950 onwards this started to change, and the German economy experienced what was called “the economic miracle”. From 1950 onwards there was also high educational expansion for both men and women. And as in most European countries, Germany has undergone rapid changes in family formation. Although at the beginning of the 20th century ages at marriage and motherhood fell, they have been increasing again since the 1960s. Age at first marriage is rising and, Germany is confronted with low fertility rates and an increasing age at entry into first motherhood across birth cohorts. Family formations that were uncommon before, like single-parent families and female-headed families have become increasingly common (Blossfeld & Jaenichen, 1992). An increasing share of women remains childless, and couple households have become less common (Hanewald & Kluge, 2014). Furthermore, divorce rates almost doubled in the 1970s (Elderly & George, 2016). These changes cannot be observed apart from changes in the educational system and the labour market. All over Europe there is a re-evaluation of the role of women in society, better career opportunities for females in the workplace and increased female labour force participation (Blossfeld & Jaenichen, 1992, p.302).

1980 was a period in which high educational expansion continued, but this was combined with growing economic uncertainty (Blossfeld & Jaenichen, 1992, p.306). Globalization processes significantly changed the way young people experienced the transition into adulthood. Internationalization of markets, intensified competition, accelerated spread of networks and knowledge through technologies, and random shocks, affect many countries around the globe. (Mills & Blossfeld, 2003). The increasing dynamics that globalization brings with it make it more difficult for individuals to predict the future and make choices between different alternatives and strategies. An increase in uncertainty about social and economic developments is therefore a defining feature of globalization (Ibid.). There are however some, as Mills and Blossfeld (2003) call it, institutional filters that channel uncertainties. Examples are the educational system, family systems, welfare regimes and employment systems. However, those institutional filters are

(13)

13 currently changing in Germany.

Although from 2015 Germany sees the highest birth-rates since 1990 it is still lagging behind most other European countries. This results in an aging population (The Local, 2017). Due to the German declining population, state-owned housing was largely sold off and almost no new houses were built. The result is that today it is incredibly challenging to find affordable housing in German cities (Pieper, 2018). Meanwhile, the welfare state is shifting its family policies from transfer-intensive to employment-oriented policy (Seeleib-Kaiser, 2016). Social rights are made conditional in line with work. While social rights are made conditional in line with work, in the past decades the German labour market has undergone profound changes. After the 1980s, unemployment rates in Germany were one of highest among rich democracies. In 2000 however, unemployment rates were running back, and scholars began talking about a new German miracle. But during this period, earnings and wages inequality were increasing and precarious employment was rising which heightened economic insecurity (Brady & Biegert, 2017). From 1990, three important structural changes that altered German labour market relations can be distinguished. The first is a sharp increase in low-paid work since 1994. On itself low-paid work does not say much about poverty because low-paid workers usually can depend on other household members. But working poverty, which is a measure of all individuals in employment households, is also rising rapidly. In 1994 working poverty rose to 2,9%, in 2000 to 3,2%, and to 4,4% in 2013. Moreover, there is a growth in temporary contracts. It is common knowledge that those with a temporary contract work on a more provisional and temporary basis, which results in more job insecurity. Temporary contracts in Germany increased from 11.5% in 1994 to 14.6% in 2013 (Brady & Biegert, 2017). Deregulations of the German labour-market also led to the rise of “mini-jobs”, whereby the employer does not have to pay social insurance taxes or offer employment protection (Ibid., p.12). These labour market changes resulted in a weakened position of workers on the labour market. Workers with temporary contracts have a higher risk of losing one’s job than workers with a permanent contract. The emergence of precarious jobs, however, implies that education no longer naturally leads to a stable job, consequently, German youth unemployment rates are at its highest (Ibid).

(14)

14 2.3 Why external events cause attitudinal change

In the previous chapter I analysed changes in German society that potentially influenced the attitudes of individuals born in different cohorts. But it is not evident that experiencing a war or an economic crisis necessarily has a long-lasting impact on risk attitudes. Therefore, two important questions should be asked; how and on which age do attitudes get shaped?

There are several theories about attitudinal development. Some theories, like the increasing persistence hypotheses state that individuals are flexible and responsive to social circumstances. This flexibility is more prominent at a young age, as people get older their flexibility gradually declines. Another theory, the life-long openness hypothesis states that individuals constantly alter their attitudes in response to changing circumstances or life events. But the most common theory about attitudinal development is the impressionable years hypothesis. This hypothesis states that individuals are especially prone to socializing influence during young adulthood, which is from around the age of 18 to the age of 30. Once socializing years have passed an individual’s attitudes are somewhat fixed and as a result, birth cohorts resemble each other in terms of basic attitudes (Krosnick & Alwin, 1989). Krosnick & Alwin (1989) use panel data to test two theories about attitudes change; the increasing persistence theory and the impressionable years theory and find evidence that points to the impressionable year’s hypothesis. Their results demonstrate a stable pattern in attitudes after the age of 30.

Roughly two processes make these impressionable years so important. The first is increased exposure to ‘change-inducing circumstances’ and the second is referred to as ‘youth susceptibility’ (Dinas, 2013). Increased exposure to change-inducing circumstance means there is a greater density of significant life events, like going to college, entering the labour market or leaving the parental house, during young adulthood and a wider spacing of such events later in life. Throughout this process, young adults are exposed to various stimuli and encounter many different new world views. But early adulthood is not only important because the amount of significant life events but also because of the weight young individuals attach to these events. Mannheim (1952, in Dinas, 2013) referred to this as ‘the stratification of experience.’ This holds that although at any given point in time all cohorts experience the same event, they respond differently according to their age. The older generation already has many experiences and collected several kinds of information therefore, a particular event constitutes only another layer in their already existing stock of information. The younger generation has a lack of prior experience which makes it

(15)

15

difficult to superimpose events on early impressions. This makes young adult more open to change (Ibid., p.868). Therefore, macro-events are especially influential for individuals who experience this in early adulthood, and from the age of thirty attitudes are mostly fixed. This attitudinal stability implies one should observe the same willingness to take risks when observed repeatedly over time. Schildberg-Hörisch (2018) used the German Socio-Economic Panel survey to test the stability in risk attitudes and concludes that risk preferences appear to be persistent and moderately stable over time. The degree of stability is however too low to conclude perfect stability.

The second important question is, how do risk attitudes get shaped? The following chapters will elaborate more on the possible mechanisms for attitudinal change, and link this to specific birth cohorts and ages to form hypotheses. If external events alter risk-attitudes in a permanent way, and attitudes towards risks influence all kinds of decision-making processes, this means that the consequences of macro-events will have a long-lasting impact on a society.

2.3.1 Traumatic experiences

Psychological shocks, such as a war, natural disasters or civil conflicts can have a deep psychological impact on people and therefore influence all kinds of human perceptions and behaviour, including the willingness to take risks (Bogdan, Just & Wansink, 2013; Deahl et al., 2000; McFarlane, 1989). The precise impact from psychological shocks on attitudes towards risks is a topic of debate. Although some researchers indicate that experiencing psychological shocks make individuals more risk averse, others find evidence that points in the opposite direction (Hanoka, Shigeoka & Watanabe, 2018). There is no sufficient explanation for the inconsistency of the results as the literature about this topic is quite young. Schildberg-Hörisch (2018) however points to two possible downfalls in the literature. The first is that most research about how natural catastrophes or violent conflict alters attitudes towards risks use experimental data from developing countries, while the research tools are mostly developed in the context of high-income countries. This might distort the results as individuals in the sample are more often less educated, partly illiterate or less used to abstract thinking. Moreover, most of the literature is suffering from a lack of theoretical predictions about the circumstances under which risk attitudes change (Schildberg-Hörisch, 2018).

(16)

16

Most scholars do however agree that the behavioural effects are caused by emotions related to mental stress. Whereas some emotions (like fear) lead to more pessimistic risk estimation which result in risk averse choices, while other emotions (like anger) lead to more optimistic risk estimations and therefore result in risk tolerant choices (Lerner & Keltner, 2001). Emotional states can alter perceptions of the surrounding. Prior research indicates that when people feel sad, they perceive a hill actually steeper than when they feel happy. This indicates that the perception of the visual surrounding is influenced by non-optical factors, including emotions (Zadra & Clore, 2011). One-way emotions alter perceptions is through attention, if an someone feels anxious, their attention will go to possible dangers. This focus on possible dangers shapes their perception; their surrounding looks more dangerous now and in turn induces emotions like stress. So, although emotions are responses to something that happens in the surrounding, it also effects the affective state of an individual. Furthermore, individuals use these affects and emotions as information about the energy costs of potential actions. Therefore, they are an important factor in decision making processes (Ibid.). Although there are some scholars who suggest that it is the other way around, and individuals in a good mood will try to protect their mood by avoiding risks an individuals in a bad mood try to uplift their mood by taking some risks, most evidence does not point this way (Schildberg-Hörisch, 2018).

The argued mechanism between emotions, or affective states, and attitudes towards risk is that anger is associated with high certainty, high control, high responsibility and medium anticipated effort while fear is associated with low certainty, low control, high anticipated effort and medium responsibility. Therefore, when a person feels anger, it produces the tendency to perceive risks as predictable and therefore controllable while fear has the opposite effect (Fessler, Pillsworth & Flamson, 2004, p2/3). Guilt and anxiety are other emotions that highly correlate with risk averse attitudes (Guiso, 2014). Thus, emotions direct attention towards certain aspects of a given situation which in turn is used as information to shape an attitude about potential risks which shapes the willingness to take those risks.

The above picture, where fear leads to risk averse and anger to risk tolerant attitudes is of course a very simplistic one. Emotions alternate and a situation never evokes just one emotion at a time. But the underlaying mechanism between attitudes towards risks and emotions is that different emotions come with different levels of (perceived) control, certainty, responsibility and anticipated effort. Thus, emotions are important to make sense of a situation or a phenomenon and

(17)

17

it is an important mechanism through which individuals come to understand themselves and their contexts. Our affective states temper or intensify our attentions, investments and attachments and help to solidify our beliefs and ideas (Gould, 2009). Furthermore, emotions are not just an individual experience, it is often difficult to make sense of these vague and bodily sensations that are called emotions. The labelling of emotions therefore brings these vague and bodily sensations into the realm of cultural meanings which structure our feelings. In the process of translating emotions into language, it becomes structured by social conventions (Ibid.). Some feelings are experienced as inappropriate in a certain situation, while other feelings, like guilt or shame, are primarily shaped through social interaction. Therefore, social conventions also structure an individual’s feelings.

Because of this cultural aspect of emotions, events can evoke shared emotions among the people who endure it. Bogdan, Just and Wansink (2013) study financial risk taking among ex-veterans, where most of their respondents were in military service during World War II. Their results indicate that combat experience significantly decreases the probability that ex-veterans hold risky assets with a decrease around 17%. The authors did not find a significant effect between combat experience and the probability of holding safe assets, which indicates that combat experience only affects the holding of risky assets. Furthermore, their results indicate that the effect of combat experience on the holding of risky assets is much bigger than that of education. (Ibid., p.461).

Other traumatic experiences, like earthquakes also have a significant effect on attitudes towards risks. But there are conflicting outcomes about the precise effect on risk attitudes. Hanoka, Shigeoka and Watanabe (2018) use panel data in which the same respondents are asked before and after the great East-Japanese earthquake. The authors conclude that men who were exposed to higher intensities of the earthquake became more risk tolerant, the magnitude of the effect remained stable during the length of the research which was five years after the earthquake. The respondents were asked about their mental state, and they reported feelings of mental stress caused by the earthquake. Their results suggest that psychological shocks can fundamentally alter risk preferences and have a long-lasting impact on risk attitudes. Their study does not separate their sample by age; therefore, it is unclear from their research if the effect of the shock differs for individuals from a younger or older age.

(18)

18

experience of a psychological shock, but in which direction the effect goes is not clear and is probably depending on the interpretation and the context of the experience.

2.3.2 Labour market changes:

The labour market in West Germany is, as it is in other European countries, rapidly changing since 1980. The question remains how economic circumstances are linked to attitudes towards risks?

An increasing risk of losing one’s job leads, according to Hetschko and Preuss (2015), to a decrease in the general willingness to take risks. Sahm (2012) on the other hand, studied the same topic but came to the opposite conclusion. The crucial difference between the two studies was whether job loss came as a surprise. In the research conducted by Sahm (2012) employees lost their job because of individual reasons, where in the study from Hetschko and Preuss (2015) the loss of employment was due to something beyond the workers power: the closing of the plant. If individuals lose their jobs because of something that is beyond their control, their perceived levels of control and responsibility, as was indicated as a mechanism through which risk attitudes get shaped in the previous chapter, would be low. The structural labour-market changes in West Germany increased the risk of losing one’s job without the worker being able to influence it and without losing your job due to individual responsibility. If we follow the reasoning of Hetschko and Preuss (2015) this would lead to a decrease in the general willingness to take risks. Although Hetschko and Preuss (2015) only study individual job loss, there is evidence that macro-economic experiences effect the individual willingness to take financial risks. Malmendier and Nagel (2011) find in their study that the birth cohort which experienced the Great Depression at a young age became more risk-averse than birth-cohorts who did not have the same experience or were in a different age category. Their results show that experiencing a financial crisis is especially influence full on young adults. Moreover, Guiso, Sapienza and Zingales (2018) conclude that the financial crisis of 2008 effected the willingness to take risks because of in increase in fears for the future.

Although it is unclear if these results also apply to general risk-attitudes. Given the rise of precarious employment, it is imaginable that economic believes about the future are less optimistic than before, which can negatively influence the general willingness to take risk. Additionally, several studies find a relationship between income and the willingness to take risks (Wright, 2017). It is generally held that there is a positive relation between income and the willingness to take

(19)

19

risks, although there is much debate about the direction of causality. Wright (2017, p.15) used a longitudinal survey to study if a high income leads to more risk tolerant attitudes. Wright (2017) carefully concludes that there is most likely a causal effect between income and an individual’s willingness to take risks.

Changing economic circumstances are especially difficult for youth entering the labour market. Individuals who already gained several years of experience or established a career have a greater chance of being hired or have enough experience to avoid dismissal (Mills & Blossfeld, 2003). In short, macro-economic events can shape individual risk-attitudes, where negative shocks (like a financial crisis) lead to more risk averse attitudes which can have a long-lasting effect.

2.3.3 Educational expansion:

Rising educational levels are a key defining feature in Europe since the 1950’s. And education might be one of the most prominent protection mechanisms for future risks, as it normally leads to higher wages and better job security. On the other hand, going to college is also taking a certain amount of risk. During the length of the study you must surrender present earnings, while borrowing governmental money to hopefully obtain a higher income in the future. The link between education and risk attitudes is already extensively studied. Establishing causality is however difficult, and there is no consensus whether education is correlated with risk aversion or risk tolerance (Jung, 2015). There are, to my knowledge, two studies that try to establish causality. The first is from Jung (2015) who uses a regression discontinuity design (a pretest-posttest design) and concludes that education increases risk aversion. This is, however, only the case with lower education, the effects are reversed for higher educated individuals; they become more risk tolerant. Chong and Martinez (2015) confirm these results and conclude that causality goes from education to risk attitudes. A possible explanation for the difference between educational levels and risk attitudes is that compulsory education may increase risk aversion by making individuals more aware of the risks around them. While tertiary education can provide with tools for managing these risks and therefore diminish risk aversion (Jung, 2015). Rising educational levels in Germany should, in line with Jung (2015) make individuals more risk tolerant. But based on the conflicting findings from researchers, this correlation might not be so straight forward. Previous research both finds a correlation between education and risk tolerance and education and risk aversion. Following the reasoning of Jung (2015) these different relations between education and risk

(20)

20

attitudes can be explained by educational levels and the tools they provide to manage certain risks. One important tool that education provides to manage risks is that education increases the chance to get a well-paid and stable job. And a well payed and stable job can provide for other things that make life less uncertain, like finding appropriate housing. Unfortunately, fi006Eding a stable job after college is not so self-evident anymore, as young adults in West Germany have experienced declining employment rates since 2004 (Brady & Biegert, 2017). This could mean that although rising educational levels increase the willingness to take risks, a changing context can diminish these results. Furthermore, education does not only provide economic tools, but it also affects many cognitive functions like analytical thinking, planning or reflection. These tools can give you the mental ability to better calculate and manage risks. This, in turn, can lead to a more optimistic risk taxation which might lead to a greater willingness to take risk.

2.3.4 Changes work-family interface

Lower marriage rates and higher divorce rates in West Germany means that individuals can rely less on intra-family risk sharing and resource pooling. Married couples benefit from economies of scale, face lower participation costs when they invest in stock markets and benefit from income-risk sharing. Therefore, being married can provide protection against some economic uncertainties which heightens the chance of financially risk-taking behaviour. Previous research indicates that married individuals are indeed more likely to invest in stock-markets and hold portfolios with higher-risk levels (Hanewald & Kluge, 2014). On the other hand, some scholars argue that married couples more often have children than unmarried couples, which place a high resource demand for their basic ‘survival needs’. This reduces resources available for risky investments (Chaulk, Johnson & Bulcroft, 2003).

Parenthood in general is related to more risk averse attitudes. Görlitz and Tamm (2015) study the general willingness to take risks with the SOEP dataset and find that both men and women are significantly more risk averse after the birth of the first child. Their risk attitudes are lowest shortly after the birth while the effect becomes weaker when the child grows older. Both parents already become more risk averse two years before the birth of the first child and twelve years after the birth of the child the effect becomes statistically insignificant. The effect from childbirth on risk attitudes is only present when individuals become parents for the first time and

(21)

21

the effects do not change due to marital status (Görlitz & Tamm, 2015, p.5). The fact that the authors only find a negative effect on risk tolerance with the birth of the first child might be (partly) explained by the fact that the second child is usually born within a couple of years after the firstborn. During this time, parents already became more risk averse due to the birth of their first child.

From the literature it seems that marriage and parenthood are related to attitudes towards risks. An important mechanism through which marriage and parenthood influence the willingness to take risks is that it alters the social context and social expectations. Marriage offers benefits to manage certain risks and mitigate other potential risks while parenthood brings with it a certain amount of responsibility which makes individuals more aware of potential risks. Furthermore, when individuals enter marriage or parenthood, a whole new set of expectations comes in play. In traditional gender roles, for example, marriage for women usually means they devote themselves to the household and withdraw from the labour market. In the last decades these traditional gender roles are changing, and female education and labour market participation become increasingly common (Chaulk Johnson & Bulcroft, 2003).

Moreover, the shift in welfare state policies place much more emphasis on work as a central pillar. West Germany’s social security system has forged an inseparable bond between employment and welfare by linking social benefits to people’s jobs (Evers & Ewert, 2012). These developments alter societal expectations and existing family roles. As women become more educated, they also want to pursue a career more often. Combined with changing welfare state policies, the labour market becomes of a more central focus in the lives of individuals.

In sum, marriage offers less protection, as marriage rates decline, and single parent households become increasingly common. An increasing central focus on work in people’s lives change societal expectations, in which the lives of both men and women are increasingly intertwined with the labour market. This entanglement of family life with the labour market, while labour market insecurities are increasing and working poverty is rising, translate in a more insecure and less predictable future.

(22)

22 2.3.5 Gender differences

Several studies indicate there are gender differences in the willingness to take risks, where women are on average more risk averse than men (Eckel & Grossman, 2002). Researchers indicate that men and women differ in their risk perceptions regarding alcohol and drug use, on the dangers of technology, industrial hazards, radioactive waste, environmental degradation and various social and recreational activities. Regarding risk taking, men are more likely to gamble, pursue ‘direct risk’ health related behaviour and take more financial risks than women (Ibid. p.2/3). Closely related, psychological research indicates that women experience stronger emotional reactions to risky situations than men. Women report more nervousness in anticipating negative outcomes than men and women tend to feel fear while men tend to feel anger when facing adverse outcomes (Chadi & Jirjahn, 2019). Because risk attitudes play such a vital role in many decision-making processes, a gender gap in risk attitudes can have far-reaching consequences for women. Women all over the world earn less than men, end up in different occupations than men and have a lower chance to end in in a managerial role. Risk attitudes, next to discrimination and a series of other factors, might be an important factor that contribute to these gender differences. A higher risk aversion in women can, for example, imply that women are less likely to negotiate with their boss for a better salary or a higher position or are less successful in these negotiations (Ibid.). From an evolutionary perspective, gender differences have evolved because men and women had different strategies to maximize the fitness of their genes. In hunter-gathering societies women were mostly focussed on parenting while men went hunting. In hunting, taking a bit of risk, like hunting for large animals could potentially bring great success and those hunters would have more and healthier offspring. While women were more invested in parenting, they were more sensitive for the possibility of loss to protect their offspring. Especially young children are sensitive to fluctuations in resources, and women were therefore more likely than men to exhibit loss aversion in behaviour (Eckel & Grossman, 2002). Other scholars argue that differences between men and women are driven by biological differences like different hormone or testosterone levels and differences in brain structure (Chadi & Jirjahn, 2019).

Several social scientists argue instead that differences in risk attitudes between men and women are not due to actual biological differences, but differences are mostly shaped by the social environment. In these theories, society has different expectations of individuals based on their sex. In traditional gender roles women are expected to nurture their family while men provide the

(23)

23

financial resources and make important family decisions. These different gender roles translate into different personality traits ascribed to men and women. Women are expected to be docile and generous while men are expected to be confident and self-assertive. Along this line of reasoning, gender differences in risk attitudes appear to be a component of socially constructed norms. Men are expected to be risk takers while women are expected to have a higher degree of risk aversion. As individuals tend to conform with what is expected from their social category, this would imply that women behave more risk averse then men (Ibid.). Chadi and Jirjahn (2019) test this hypothesis with data from the SOEP in which the authors make a distinction between West and East Germany. East Germany is characterized by more equal gender roles while West Germany is characterized with more traditional gender roles. These differences between East and West Germany translate into higher employment rates among East German women and their labour force participation is little influenced by their marital status. Partnership and fertility patterns also differ between East and West Germany where East German women are emotionally and economically less dependent on a male partner (Ibid.). The authors find there is indeed a statistically significant difference in risk attitudes between women from East and West Germany. East German women are more risk tolerant than West German women. This supports the view that risk-attitudes are in fact, at least partly, shaped by society.

If differences in risk attitudes between men and women are indeed influenced by gender roles, these differences will become smaller over time. Since the 1960s a ‘sexual revolution’ has been set in motion that has changed existing gender roles. Female education was expanding, and it became less obvious that women put their career aside for their children. The family and school system are two main pillars that play an important part in gender role socialization and it takes primarily place during early childhood. Girls and boys spend most of their time at home where they look to their parents or older siblings for guidance. Parents, therefore, provide children with their first lessons about gender through role modelling and the encouragement of different types of behaviour in their sons or daughters (Leaper, 2014). When children eventually attend school, these attitudes and behaviours are reinforced or challenged by the peers of the child and the school experience (Molla, 2016).

(24)

24 2.4 Hypotheses

Based on the previous chapters, in this chapter hypothesises are formulated that help study generational change in West Germany more precisely. To study change in the willingness to take risks between different West German birth cohorts, several macro events are used as indicators. The indicators are World War II, labour market changes, educational expansion and changes in the family-work interface. Although most research indicate a relation between these variables and risk attitudes, the results are conflicting about the direction; do these events lead to risk tolerant or risk averse attitudes? These differences might be explained by the fact that these separate spheres are in fact connected. Studying one effect without taking this interconnectedness into account might result in conflicting outcomes. Although most scholars control for socio-demographic variables (like being married, employment or educational level) they do not study the relation between these spheres. This research takes a different approach, variables, like education, income or marital status are not included in the model to study their individual effect. Instead it takes a more holistic approach to study how the general willingness to take risks is changing over time in the West German society. This means that the macro events, used as indicators for change, are not tested directly.

The first hypothesis is built around World War II which can be considered a traumatic experience for the West German population. The war had a disrupting impact on many aspects of everyday life. Uncertainty became a defining feature of life in West Germany during and shortly after the war. Many aspects of life became hard to control and a sentiment of guilt was felt under the West German population. These circumstances would be especially impactful on the attitudes of West German individuals who were at an impressionable age (18- 30). Consequently, individuals who were born in the cohort of 1920 and the cohort of 1930 would be most influenced by the war and its aftermath and are therefore more risk averse than the younger cohorts. There is, however, some difference between these cohorts as the cohort of 1930 came of age during the aftermath of the war but also during the prosperous 1950’s. This might make them more willing to take risks than the cohort of 1930. Therefore, the first hypothesis is as follows;

H1: Individuals born in the cohort of 1920 and 1930 report least willing to take risks of all cohorts. This effect is strongest for the cohort of 1920.

(25)

25

Next to World War II, macro events used as indicators for cohort change in risk attitudes are educational expansion, labour market changes and changes in the family-work interface. Since the 1950s the West German economy started to flourish, and educational levels were rising. Consequently, individuals born in the cohort of 1940, 1950 and 1960 had better chances on proper education and entered the labour market in a time in which finding a job was relatively easy. Those birth cohorts would not differ much from each other in their willingness to take risks, but they would differentiate from the cohorts of 1920 and 1930. Therefore, the second hypothesis is as follows:

H2: Individuals born in the cohort of 1940, 1950 and 1960 report more willing to take risks than individuals born in the cohort of 1920 and 1930. The reported change in willingness to take risks between successive cohorts is strongest between the two oldest cohorts and the cohort of 1940. From the 1980s onwards the situation changed. Economic and social insecurities were rising while education brought less benefits than before. Furthermore, family life becomes more in line with work while marriage offers less protection, as marriage rates are declining, and single parent household become increasingly common. Therefore, the third hypothesis is as follows:

H3: From the cohort of 1970 onwards, the general willingness to take risks is decreasing with every following cohort.

The fourth hypothesis is to study differences between men and women. The literature indicates that women are on average more risk averse than men. Chadi and Jirjahn (2019) argue that those differences are shaped by gender role expectations. Gender roles are changing in German society, when Hitler was in power, a strong believe was placed on traditional gender roles, with a male breadwinner and females in the role of housewives. Internalizing these social roles and socializing this to their children would mean that women born in 1920 are more risk than the women who grew up in a more equal society with less strong gender roles. From 1960 onwards, traditional gender roles were changing. Consequently, differences in risk attitudes between men and women should decline in time. Women who are born in the cohort of 1950 grew up in a more gender equal society which affects their risk attitudes. Thus, differences between men and women should become smaller with every successive cohort from the cohort of 1950 onwards.

H4: Women are on average more risk averse than men. Differences between men and women decline with every following cohort from the cohort of 1950 onwards.

(26)

26 3. Research Methodology

3.1 Data and sample

Within this paper panel data from the German Socio-Economic Panel survey from 2016 will be used to answer the main research question. The German Socio-Economic Panel survey is a longitudinal study of private households. Every year over 30.000 respondents in nearly 11.000 households are interviewed on a wide range of topics such as attitudes and values, household composition, work, education and social networks. The study was launched in 1984 and runs op to 2016. Observations regarding the general willingness to take risks have been included since 2004, then again in 2006 and from 2008 it is asked every year (SOEP, 2019). This means that the available data covers a time period of 13 years in which individuals are studied at overlapping ages. It has been described as the “longest-term and most complete multi-cohort dataset available to model risk-taking trajectories across the lifespan using within-person data” (Josef, Richter, Samanez-Larking, Wagner, Hertwig & Mata, 2016, p.14). Because the same private households, individuals and families are questioned over a long period of time, it seizes a rare opportunity to study societal change in the willingness to take risks under the West German society.

The initial sample contained 84 862 respondents measured over 33 waves, which results in 619,718 observations. For the target sample, several cuts were made. The sizes of the cuts and their corresponding exclusion criteria are presented in Table 1. For the first sample selection, the survey years that the dependent variable was not part of the questionnaire were deleted. This resulted in an exclusion of 21 455 respondents. The second sample selection consists of an age range. Respondents were eligible to participate in the survey from the age of 17 and respondents older than 87 were excluded from the data due to their small sample size and their potentially deviant answers due to the circumstances of their old age. Another 312 respondents were excluded. Respondents with a migration background were excluded because they grew up in a different social context, possibly affecting their willingness to take risks in a way that is not representative of the German population. Another 21 178 respondents were excluded. To increase the overall sample size, the SOEP has added several refreshment samples (1998, 2000, 2006, 2011 and 2012). The SOEP also included several other group samples, like high income or single-parent households (SOEP, 2019). The refreshment samples are included in the target sample. The group samples are excluded to prevent a bias towards certain groups, since these households live in different

(27)

27

circumstances which might influence their risk attitudes. With the exclusion of the group samples another 17 770 respondents were excluded.

The initial plan was to include both West and East Germans in the target sample to make a comparison between the two. East and West Germany were, and on some level still are, two completely different worlds after World War II. Therefore, many of the external events that can influence risk attitudes differ between the two parts of Germany. Plotting the willingness to take risks from East and West German individuals indicates there are indeed important difference (see Appendix I). Although it might be interesting to study the causes of these differences, it wanders to far from the purpose of this study. Therefore, I decided to exclude East Germans from the target sample by deleting respondents who answered “East Berlin” on the question where they lived in 1989 (deleted N= 4 299). To still capture the youngest cohort (1990) respondents who answered “West Berlin” on that same question were included in the target sample. As follows, the cohort of 1990 is born in a unified Germany. Individuals who answered “Abroad” on the same question were excluded (deleted N= 53).

After these selections the sample consisted of 108 390 observations and 19 795 distinct respondents. Next, a categorical cohort variable was created starting at the year 1920 and cut every ten year. Respondents before 1920 were dropped from the sample due to their small cohort size. Therefore, 67 respondents were excluded. Another 24 respondents were excluded because they were missing on the dependent variable. The final sample consists of 10 166 men and 95 38 women (N = 19 704) born in West or a unified Germany between 1920 and 1998 in the age between 17 and 87. The total amount of observation is 107 927.

(28)

28 Table 1: Sample selection

N (number of respondents cut) Percentage

Starting sample 84 862 100

Exclusion criteria Outcome not part of

questionnaire

21 455 25.3%

Younger than 17 and older than 87 312 0.37% Migration background 21 178 25% Refreshment samples 17 770 21% East German 4 299 5.1% Living Abroad in 1989 53 0.06% 1910 birth cohort 67 0.08% Missing on outcome variable 24 0.02% Selected Sample 19 704 23.2%

Source: German Socio-Economic Panel (SOEP), version 33, release 2017

3.2 Operationalization and validity

To measure general attitude towards risk taking of the West German society, the following question will be used: “Are you generally a person who is willing to take risks or do you try to avoid taking risks? Please tick a box on the scale, where the value 0 means not at all willing to take risks and the value 10 means very willing to take risks.” (SOEP, 2019). This question was asked 11 times starting in 2004, then again in 2006 and from 2008 onwards it was asked every year. Table 2 provides an overview of the descriptive statistics of the variables used in this paper. The table shows that the mean value of the willingness to take risks is 4.49 with a standard deviation of 2.32.

(29)

29

There are, however, some potential problems with analysing attitudes towards risk taking with a survey question. Whether survey questions are a good method for measuring risk attitudes is a point of discussions amongst scholars. Some scholars are sceptical about whether this self-reported willingness to take risks is behaviourally meaningful (Dohmen et al., 2011). Dohmen et al. (2005, p.3) try to overcome this problem by testing the validity of the general risk-taking question in the SOEP by conducting a field experiment. Participants were asked to answer the general risk-taking question and participate in an experiment in which they had to make risky choices with real money at stake. The authors used a representative sample of 450 German adults and found that the answer to the general risk-taking question was a good predictor of the actual risk-taking behaviour in the experiment. In addition, Dohmen et al. (2005) also compared the answers of the general risk-taking question with a more standard lottery question which is available in the SOEP. This lottery question gives individuals a hypothetical investment opportunity: respondents were asked how much of a windfall gain of 100.000 euros they would invest in an asset that would return half, or double the amount of money, where the probabilities are equal. The benefit of this question is that it gives clear probabilities, therefore holding perceptions of the riskiness of the situation constant for all respondents. Furthermore, risks are in this situation separated from uncertainty. It turns out that the invested amount of money in the hypothetical investment situation is strongly correlated with responses to the general risk-taking question (Ibid., p.4). Moreover, the general risk question remained a reliable predictor of actual risk-taking behaviour, even after controlling for a large number of demographic and economic variables (Dohmen et al. 2011).

Another potential difficulty is whether attitudes towards risk taking differ between different context, like health risks, financial risks or social risks. With the general risk-taking question this nuance is lost. In the SOEP several questions distinguish between the willingness to take risks in these separate domains. Dohmen et al. (2005) studied the correlation between the different questions regarding risk taking and their results indicate that although the willingness to take risks differs across contexts, the correlation between the questions is high. Principal component analysis tells the same story: there is one underlying component which explains the variation (Ibid.). Therefore, it is safe to say that although there are some contextual differences, the general risk-taking question is a good indicator for general attitudes towards risk risk-taking.

(30)

30 Table 2: Descriptive statistics

Variables N (obs.) Range Mean SD Description

General risk attitudes

107 927 0 - 10 4.49 2.32 Ranging from:

0 = not at all willing to take risks, to

10= very willing to take risks Age 107 927 17 - 87 51.57 17.74 Survey year minus birth year

Sex 107 927 0 - 1 0.521 0.50

Survey year 109 727 2004 - 2016

Birth cohorts 109 727 1920 - 1989 Birth years combined into 10-year intervals

Cohorts 109 727 Average willingness to take

risks per cohort

1920 – 1929 3768 3.21 2.44 1930 – 1939 13931 3.75 2.36 1940 – 1949 18135 4.26 2.31 1950 – 1959 18981 4.42 2.28 1960 – 1969 23384 4.64 2.21 1970 – 1979 13006 4.69 2.19 1980 – 1989 10575 5.15 2.23 1990 - 1999 6142 5.65 2.18

Are you willing to take a risk? : a panel study of cohort change in risk attitudes in West Germany 1984-2016