Household income expectations: The role of unexpected income changes and aggregate conditions

Bucciol, Alessandro; Easaw, Joshy Z.; Trucchi, Serena Working Paper Household income expectations: The role of unexpected income changes and aggregate conditions MUNI ECON Working Paper, No. 2025-04 Provided in Cooperation with: Masaryk University, Faculty of Economics and Administration Suggested Citation: Bucciol, Alessandro; Easaw, Joshy Z.; Trucchi, Serena (2025) : Household income expectations: The role of unexpected income changes and aggregate conditions, MUNI ECON Working Paper, No. 2025-04, Masaryk University, Faculty of Economics and Administration, Brno, https://doi.org/10.5817/WP_MUNI_ECON_2025-04 This Version is available at: https://hdl.handle.net/10419/324767 Standard-Nutzungsbedingungen: Die Dokumente auf EconStor dürfen zu eigenen wissenschaftlichen Zwecken und zum Privatgebrauch gespeichert und kopiert werden. Sie dürfen die Dokumente nicht für öffentliche oder kommerzielle Zwecke vervielfältigen, öffentlich ausstellen, öffentlich zugänglich machen, vertreiben oder anderweitig nutzen. Sofern die Verfasser die Dokumente unter Open-Content-Lizenzen (insbesondere CC-Lizenzen) zur Verfügung gestellt haben sollten, gelten abweichend von diesen Nutzungsbedingungen die in der dort genannten Lizenz gewährten Nutzungsrechte. Terms of use: Documents in EconStor may be saved and copied for your personal and scholarly purposes. You are not to copy documents for public or commercial purposes, to exhibit the documents publicly, to make them publicly available on the internet, or to distribute or otherwise use the documents in public. If the documents have been made available under an Open Content Licence (especially Creative Commons Licences), you may exercise further usage rights as specified in the indicated licence. https://creativecommons.org/licenses/by/4.0/ n. 2025-04 ISSN 2571-130X DOI: 10.5817/WP_MUNI_ECON_2025-04 Published in Journal of Economic Behavior & Organization, 2025 (Article) Household Income Expectations: The Role of Unexpected Income Changes and Aggregate Conditions Alessandro Bucciol / University of Verona, Department of Economics Joshy Easaw / Cardiff University - Cardiff Business School Serena Trucchi / Cardiff University - Cardiff Business School, Netspar and Masaryk University Household Income Expectations: The Role of Unexpected Income Changes and Aggregate Conditions Abstract We analyse how unexpected income changes and aggregate conditions influence income expectations, their uncertainty, and expectation errors. We use a uniquely rich longitudinal Dutch survey collecting detailed information on the distribution of household income expectations. Our results show that unexpected income changes, much more than aggregate conditions, induce a revision in income expectations across the entire spectrum of the expected income distribution, consistent with extrapolative behaviour. We also document that unexpected income changes increase the uncertainty about future income. Our results provide some evidence of over-reaction, particularly to negative unexpected income changes and among high-income individuals. These effects differ based on an individual’s position in the income distribution, which may be attributed to differences in income dynamics, role of insurance mechanisms, and varying levels of awareness about how unexpected income changes and aggregate conditions impact household finances. Masaryk University Faculty of Economics and Administration Authors: Alessandro Bucciol / University of Verona, Department of Economics Joshy Easaw / Cardiff University - Cardiff Business School Serena Trucchi (ORCID: 0000-0002-1932-7201) / Cardiff University - Cardiff Business School, Netspar and Masaryk University Contact: [email protected] Creation date: 2025-05 Revision date: 2025-06 Keywords: Income expectations, Expectation uncertainty, Expectation error, unexpected income changes, Aggregate conditions JEL classification: D84, G50 Citation: Bucciol, A., Easaw, J., Trucchi, S. (2025). Household Income Expectations: The Role of Unexpected Income Changes and Aggregate Conditions. MUNI ECON Working Paper n. 2025-04. Brno: Masaryk University. https://doi.org/10.5817/WP_MUNI_ECON_2025-04 (https://creativecommons.org/licenses/by/4.0/) Licensing of the final text published in the journal is in no way conditional on this working paper licence. MUNI ECON Working Paper n. 2025-04 ISSN 2571-130X Household Income Expectations: The Role of Unexpected Income Changes and Aggregate Conditions ∗ Alessandro Bucciol, Joshy Easaw, Serena Trucchi May 8, 2025 Abstract We analyse how unexpected income changes and aggregate conditions influence income expectations, their uncertainty, and expectation errors. We use a uniquely rich longitudinal Dutch survey collecting detailed information on the distribution of household income expectations. Our results show that unexpected income changes, much more than aggregate conditions, induce a revision in income expectations across the entire spectrum of the expected income distribution, consistent with extrapolative behaviour. We also document that unexpected income changes increase the uncertainty about future income. Our results provide some evidence of over-reaction, particularly to negative unexpected income changes and among highincome individuals. These effects differ based on an individual’s position in the income distribution, which may be attributed to differences in income dynamics, role of insurance mechanisms, and varying levels of awareness about how unexpected income changes and aggregate conditions impact household finances. Keywords: Income expectations; Expectation uncertainty; Expectation error; unexpected income changes; Aggregate conditions. JEL Classification: D84, G50. ∗We thank Tommaso Reggiani, the participants to the 2024 Household Finance Workshop in Turin, the 2023 Sheffield Household Finance Workshop, the ICEEE 2023 conference in Cagliari, the 2021 Anglo-French-Italian Macro Workshop in Venice, and seminar at Cardiff University for useful comments. The usual disclaimers apply. Corresponding author: Serena Trucchi, Cardiff University - Cardiff Business School, Netspar and Masaryk University; address: Aberconway Building, Colum Road, Cardiff, CF10 3EU (UK); email address: [email protected]. Alessandro Bucciol: University of Verona, Department of Economics; address: via Cantarane 24, 37129 Verona (Italy); email address: alessan- [email protected]. Joshy Easaw: Cardiff University - Cardiff Business School; address: Aberconway Building, Colum Road, Cardiff, CF10 3EU (UK); email address: easa[email protected]. 1. Introduction Household expectations about future income and its uncertainty are key factors in economic decision making. Life cycle models of consumption behaviour predict that higher expected income leads to increased consumption, while greater uncertainty encourages precautionary savings and reduces current consumption (Coibion et al.,2024;Jappelli and Pistaferri,2017). This, in turn, influences economic behaviour in other areas such as portfolio allocation (Fagereng et al.,2018), labour supply (Rossi and Trucchi,2016) and human capital (Patnaik et al.,2022). Expectations and consumption dynamics have broader macroeconomic implications, influencing the effectiveness and consequences of fiscal and monetary policy interventions and shaping business cycle fluctuations (recent examples are Bordalo et al.,2022;D’Acunto et al.,2024). Despite the crucial role of households’ income expectations, empirical evidence on their determinants is rare, possibly due to the limited availability of surveys collecting precise information on expectations over extended periods. This study provides new evidence on the process of expectation formation, focusing on how several aspects of income expectations – the expected value of household income, its dispersion and the expectation error – respond to macroeconomic conditions and unexpected changes in household income. Moreover, we explore how the response to income expectations varies across the income distribution. This allows us to highlight potential heterogeneity in our results, which may arise from differences in income processes or the role of insurance mechanisms, such as unemployment benefits. In addition, we can identify who is most exposed to the welfare consequences of expectation errors. The response of income expectations to unexpected income changes depends on the degree of persistence of income over time. Future income is not affected by transitory income shocks, while it reflects permanent or persistent income changes. However, individuals may have distorted expectations about the persistence of their income. In this case, income expectations may over-react to unexpected income changes, resulting in an expectation error (Massenot and Pettinicchi,2019;Cocco et al.,2022;Rozsypal and Schlafmann,2023;D’Acunto et al.,2024). After an improvement in their financial situation, individuals could overestimate future income, displaying overextrapolative expectations based on recent experience. According to the model of diagnostic expectations (Gennaioli and Shleifer,2010;Bordalo et al.,2018,2019), expectations respond to the news by overweighting future outcomes that become more likely in light of the current news. This leads to an overestimation of the likelihood of positive future scenarios following favorable news and a pessimistic bias in response to negative news. Our study investigates the relevance of diagnostic expectations in the context of household income expectations. The 2 effect of aggregate conditions on income expectations reflects the individual assessment of these conditions, the awareness of their impact on household financial conditions, and the correlation of household income with the macroeconomy and the business cycle. These relationships may vary across the income distribution and can be influenced by private or public insurance mechanisms. We use a uniquely rich dataset, the DNB Household Survey (DHS), collecting detailed data on household income expectations and realizations for a longitudinal sample of Dutch individuals. This enables us to precisely measure the magnitude of experienced unexpected income changes, defined as the deviation between actual household income and prior expectations. This represents a contribution to the literature that has largely examined the effects of income changes without distinguishing between anticipated and unexpected changes. We integrate this dataset with aggregate indicators on the unemployment rate and economic policy uncertainty to capture aggregate conditions. A unique feature of our empirical analysis is the availability of precise measures that capture the full distribution of income expectations. Unlike previous literature (Brown and Taylor,2006;Massenot and Pettinicchi,2019;Cocco et al.,2022), we are able to quantify the magnitude of revision of income expectations and not only the expected direction of the income change (improvement or deterioration).1Furthermore, by analysing the lower and upper bounds of expected income, we can also detect potential changes in the distribution of income expectations. We examine whether an increase (or decrease) in the expected value of income results from a parallel shift in the distribution, affecting both the upper and lower bounds equally, or if it is driven by a relatively larger change in either the left or right tail of the distribution. Perceived income uncertainty is measured using indicators that capture the dispersion of individual income expectations. This study is one of the first to explore the determinants of perceived income uncertainty of individuals.2The longitudinal structure of the dataset also allows to compare ex-post income realization with their expectations to precisely measure the expectation error. This comparison helps determine whether the response to income expectations reflects actual changes in individual circumstances or whether it results from an over-reaction or under-reaction to those changes. Our findings indicate that unexpected changes in household income have a significant and relevant impact on expectations, while aggregate conditions play a minor and mostly 1The only study measuring the deviation between income expectations and realizations is the working paper by D’Acunto et al. (2024), which focuses on Chinese households in the post-Covid-19 period. 2A notable attempt is Cocco et al. (2022), though their analysis is constrained by data limitations, as it only captures the direction of expected income changes. Similarly, D’Acunto et al. (2024) examine the link between income shocks and uncertainty, but their study focuses specifically on the post-Covid-19 period in China and does not account for aggregate conditions. 3 insignificant role. On average, both positive and negative unexpected changes in household income, particularly relatively large ones, prompt a revision in income expectations. Individuals experiencing an unexpected positive income change revise their expectations upward, while they revise income expectations downwards when hit by unexpected negative income changes. Around 20% of the unexpected income change are perceived as persistent: a 10% increase in the positive (negative) unexpected income changes determine an upward (downward) revision in income by 2% (2.5%). We also detect heterogeneity across the income distribution, with unexpected positive income changes being more relevant at the bottom of the distribution and negative ones at the top. Perceived income uncertainty increases with unexpected income changes among bottomand middle-income earners. In contrast, for high-income individuals, perceived uncertainty slightly increases with adverse aggregate conditions. Overall, this heterogeneity may arise from differences in income processes across the distribution, the varying relevance of insurance mechanisms, or different levels of awareness regarding how unexpected income changes and aggregate conditions impact household financial conditions. Understanding whether expectations revision reflects an over-reaction to unexpected income changes and aggregate conditions has relevant implications for individual welfare and macroeconomic outcomes. According to the life-cycle model, an unexpected income change determines a revision in optimal consumption. If unexpected income changes trigger an over-reaction in income expectations, consumers may deviate from their optimal consumption path—spending less (or more) than optimal in response to negative (or positive) income shocks. This has a detrimental effect on the ability to smooth consumption and may amplify the contraction in aggregate consumption during recessions. By comparing income expectations and their future realizations we find that revision in expectations is partly due to an over-reaction to unexpected income changes, especially for unexpected negative income changes and among high-income individuals. Top-income individuals are characterized by a lower marginal propensity to consume and larger buffer stocks. Therefore, over-reaction to unexpected income changes is mostly concentrated in the group where the consequences of sub-optimal consumption path are less severe. This study contributes to the literature that investigates the role of individual experience (Malmendier and Nagel,2011,2016;Massenot and Pettinicchi,2019;Kuchler and Zafar,2019;Cocco et al.,2022;Rozsypal and Schlafmann,2023;D’Acunto et al.,2024) and aggregate conditions (Bloom,2009;Malmendier and Nagel,2011;Coibion et al.,2021; Easaw and Grimme,2024) in shaping individual behaviour and expectations. Most of these studies focus either on individual behaviour and attitudes or on expectations about macroeconomic factors. We contribute to this literature by evaluating how the distribution of expectations about household income and its uncertainty respond to unexpected 4 income changes and aggregate conditions. The remainder of the paper is organized as follows. Section 2reviews the related literature; Section 3illustrates the data; Section 4discusses the empirical methods and results; finally, Section 5concludes. 2. Theoretical framework and literature review The general theoretical framework underpinning our analysis is based on the cognitive processes that drive expectation formation. Gennaioli and Shleifer (2010) and Bordalo et al. (2018,2019) develop a model of diagnostic expectations, in which expectations overweight future outcomes that become more likely in light of the current news. Therefore, favourable news leads individuals to overestimate the probability of positive future outcomes, while negative events cause them to overestimate the likelihood of negative future outcomes. In our specific context, this implies that there is a link between current unexpected income changes and the revision of expectations and the expectation error. Diagnostic expectations embed extrapolation. However, unlike mechanical extrapolation based on adaptive expectations, diagnostic expectations are forward-looking. Distortions arise when news provides informative insights into future events.3 A revision in income expectation following an unexpected income change may be driven by truly persistent shocks. However, if unexpected income changes are significantly correlated with expectation errors, this can indicate distorted expectations. Massenot and Pettinicchi (2019) illustrate this aspect, building on the concepts of extrapolation and over-extrapolation. If individuals consider unexpected income changes to be persistent and extrapolate their recent experience, the relationship between current and expected income growth is positive. In contrast, if they expect transitory unexpected income changes and mean reversion, this relationship is negative. Individuals over-extrapolate when they consider their income growth to be more persistent than it actually is, thus generating an expectation error. Individuals overestimate their future income following an unexpected positive income change, and underestimate it following a negative one. Similarly, Rozsypal and Schlafmann (2023) illustrate an expectation formation rule based on the over-persistence bias, where individuals overestimate the persistence of their income process.4The main difference between Rozsypal and Schlafmann (2023) and 3Another important factor influencing how individuals form their expectations is cognitive uncertainty. Enke and Graeber (2023) show that individuals who report higher levels of (subjective) cognitive uncertainty tend to compress their probabilistic distributions. Consequently, they overestimate the probability of unlikely events while underestimating the probability of likely ones. 4Rozsypal and Schlafmann (2023) model expectation formation in the context of a standard income 5 the diagnostic expectations approach of Gennaioli and Shleifer (2010) and Bordalo et al. (2018,2019) is that in the latter the expectation error depends on the latest news, whereas in Rozsypal and Schlafmann (2023) it depends on the history of individual unexpected income changes. Studying the response of income expectations and expectation error to new information, specifically aggregate conditions and unexpected income changes, this paper contributes to this literature by empirically examining the relevance of diagnostic expectations and its heterogeneity across the income distribution. A critical aspect of our analysis is the inclusion of a measure of unexpected income changes, rather than just income changes. This is crucial as unexpected income changes represent an update to an individual’s information set, providing a more nuanced understanding of the cognitive processes involved. By analysing the effect of unexpected changes in household income and aggregate conditions on individual income expectations, this paper builds on the empirical literature studying the effect of experiences on economic outcomes. These studies consider either the role of macroeconomic conditions experienced during the life-cycle and in the recent past (Malmendier and Nagel,2016;Kuchler and Zafar,2019) or the role of personal experience and individual events (Bucciol and Zarri,2015;Bucciol and Miniaci,2018;Cocco et al., 2022;Rozsypal and Schlafmann,2023;D’Acunto et al.,2024). The first group of studies examine whether people living through different macroeconomic histories differ in their expectations, attitudes and behaviour. Risk attitudes, expectations and portfolio composition are influenced by experiences of stock market returns and economic depression (Malmendier and Nagel,2011;Guiso et al.,2018;Angelini and Ferrari,2021;Heiss et al.,2022) and high inflation (Malmendier and Nagel,2016; Malmendier and Botsch,2020;Malmendier and Wellsjo,2024). These studies provide evidence that aggregate experience affects economic expectations, with a primary focus on expectations of macroeconomic variables, such as inflation or stock market trends. We add to this recent literature by linking aggregate experience with expectations of individual outcomes, namely future household income. In doing this, we also focus on Roth and Wohlfart (2020), who show how individuals’ macroeconomic expectations affect their personal economic prospects. Personal events have also been shown to have a relevant impact on individual attitudes, behaviour and expectations. For example, personal experience with portfolio risks and returns (Kautsia and Knupfer,2008;Bucciol and Miniaci,2018), life-course negative events (Bucciol and Zarri,2015), and a natural disaster (Hanaoka et al.,2018) influence financial risk propensity and risk-taking. Our approach is related to these studies in that process with permanent and transitory unexpected income changes, while Massenot and Pettinicchi (2019) do not explicitly model the income process but they assume an AR(1) process for income growth. 6 For each dependent variable, standard statistical tests find the fixed-effect model to describe the data better than the pooled model (without individual fixed effects) and random-effect model (where individual effects are absorbed in the error term); results are available upon request. In what follows, we adopt the convention to comment on coefficients significant at least at the 5% level. 4.1. Benchmark results Table 2outlines the results of the benchmark analysis. In general, unexpected changes in household income play a more relevant role compared to aggregate conditions, which only marginally affect all the measures of income expectations we analyse. Looking at income expectations, results in Column 1 show a significant effect of both positive and negative unexpected income changes. Positive (negative) unexpected changes increase (decrease) expected income, consistent with extrapolative behaviour, as in Massenot and Pettinicchi (2019); D’Acunto et al. (2024) and Cocco et al. (2022). The effects are similar in magnitude: a 10% unexpected income change leads to a revision of 2% for positive changes and 2.5% for negative ones. This suggests that individuals perceive 20–25% of unexpected income changes as persistent. These revisions impact the entire distribution of expectations, as shown in Columns 2 and 3. Positive unexpected income changes increase both the minimum and the maximum expected income, and negative unexpected income changes decrease both bounds. The effects of positive and negative changes are symmetric, with unexpected positive income change having a greater impact on the upper bound of income expectations and unexpected negative income changes on the lower bound. By widening the spread between upper and lower bounds, both positive and negative unexpected income changes affect the perception of income uncertainty (Column 4). A 10% unexpected income change leads to an increase in perceived uncertainty of 0.5-0.6%. This result is in contrast to evidence in Cocco et al. (2022), showing an increase in expectation dispersion only following a deterioration in financial conditions. This difference could be attributed to the explanatory variables used: we identify unexpected income changes, while Cocco et al. (2022) focus on changes in financial conditions, which can be either unexpected or anticipated. The effect of unexpected income changes on the standard deviation of expectations (Column 5) is less significant and smaller in magnitude. Focusing on aggregate conditions, unemployment significantly increases uncertainty, but its effect is small, consistent with firms uncertainty measures (Easaw and Grimme,2024). Hence, an increase of 1 percentage point in the unemployment rate results in an increase in the standard deviation by 0.2%. On average, economic policy uncertainty does not 13 significantly affect the perception of income uncertainty.12 We examine expectation errors (Column 6) and their magnitude (Column 7) to assess whether expectations reflect actual income realization or if they overreact to unexpected income changes, in line with over-extrapolation (Massenot and Pettinicchi,2019;Cocco et al.,2022;D’Acunto et al.,2024) and diagnostic expectations (Bordalo et al.,2018,2019). Expectation errors, defined as the difference between the ex-post income realization and its expected value in the previous period (Exp. errt=yt+1 −Et[yt+1]), are unbiased, as indicated by the non-significant constant in Column 6. Unexpected income changes significantly alter expectation errors, with unexpected negative income changes having more than twice the impact of positive ones. Specifically, a 10% increase in unexpected positive income change reduces errors by 1.6%, while the same increase in unexpected negative income changes increase errors by 4%.13 The reduction in the expectation error following an increase in the unexpected positive income change (Column 6) may depend on either an increase in overforecasting, namely an increase in the size of the error when positive, or a reduction in underforecasting, namely a reduction in the error when negative. Similar argument applies to the effect of unexpected negative income changes. To disentangle these two mechanisms, we examine the absolute value of the expectation error (Column 7). The negative and statistically significant impact of an unexpected positive income change indicates an average reduction in its size, suggesting that the predominant channel is the weakening of underforecasting. On average, unexpected negative income changes increase expectation errors (Column 6), but do not significantly affect the size of the expectation error (Column 7). This indicates that unexpected negative income changes trigger both mechanisms, with some individuals decreasing overforecasting and others increasing underforecasting. These findings partly confirm the role of over-extrapolation and diagnostic expectations in explaining the response of income expectations to unexpected income changes. On average, individuals tend to reduce the size of the expectation error following unexpected positive income changes, denoting improved accuracy and the absence of over-extrapolation. However, we find evidence of over-extrapolation following unexpected negative income changes, this increasing underforecasting. The analysis of heterogeneity across the income distribution illustrated in Section 4.3 will provide further insights into these results. TABLE 2 ABOUT HERE 12These findings confirm the results reported by Piccillo and Poonpakdee (2021). When examining a similar time frame, they find a statistically insignificant relationship between economic policy uncertainty and subjective income uncertainty. 13The heterogeneity in the impact of unexpected negative income changes on expectation errors is primarily driven by top-income earners, as shown in Table 3and discussed below. 14 4.2. Sensitivity and robustness checks In the appendices, we check the sensitivity and robustness of our results using alternative sample restrictions and specifications. In Appendix Awe show results on alternative samples. We enlarge the sample and include partners and respondents who report income bands for household income. Our results are also robust to omitted variables according to the Oster (2019) test; see Appendix B. In Appendix C, we study the robustness of our findings to changes in the specification. We consider six cases. In Appendix Table C.1 we replace unexpected income changes with an “objective” measure of unexpected income changes obtained following D’Acunto et al. (2024). The objective measure is obtained as the residual from a regression of realized income on its lagged value, plus socio-demographic controls and time fixed effects. The regression is estimated separately for four groups defined according to two dimensions: gender (male/female) and education (college degree/lower degree). We do this because unexpected income changes and subjective expectation errors might be mechanically correlated due to serial correlation in expectation errors. The estimation results largely confirm our findings, reinforcing the robustness of our analysis. In particular, similar to D’Acunto et al. (2024), we find an over-reaction to income changes. In Appendix Table C.2 we add to the specification two additional macroeconomic indicators, namely the inflation rate (based on the consumer price index) and quarterly GDP (in real terms, seasonally adjusted, and transformed using the inverse hyperbolic sine). In Appendix Table C.3 we replace our macroeconomic measures with year dummies to capture business cycle effects. The estimation results reinforce the limited role of aggregate economic conditions while confirming the significant impact of unexpected income changes. In Appendix Table C.4 we include in the specification a dummy equal to one for positive unexpected income changes and equal to zero otherwise. In Appendix Table C.5 we add the same dummy as in the previous exercise and dummy variables for large positive and negative unexpected income changes, alone and interacted with the size of income changes. We define “large” unexpected income changes as changes larger than the median. In this way, we investigate the heterogeneity of the effect of unexpected income changes due to their size. A graphical representation of the marginal effect of the four types of unexpected income changes on the outcome variables is shown in Figure 4. As a general result, our findings are driven primarily by large unexpected income changes. In Appendix Table C.6 we include in the specification the lagged value of the positive and negative unexpected income changes. The purpose is to assess whether the information from the most recent period is the main driver of households’ current expectations. 15 Even if we experience a drop in the sample size due to the inclusion of a lagged variable, the estimated effect of unexpected income changes remains largely significant. We find a significant impact of previous unexpected income changes on the level of income expectations (including lower and upper bounds) and the forecast error, while their effect on perceived uncertainty is not significant at standard levels. However, the effect of lagged unexpected income changes is smaller in magnitude, indicating that more recent information is more important. This finding is in line with Rozsypal and Schlafmann (2023), which shows that the entire history of unexpected income changes contributes to the formation of expectations. FIGURE 4 ABOUT HERE 4.3. Heterogeneity by income group In this section we investigate how baseline results are heterogeneous between income subgroups, identified using average household income during the observed period.14 This may contribute to understanding the drivers behind the results in Table 2and gauge their implications. There may be several factors contributing to heterogeneity. Income processes may vary across income groups, exhibiting different degrees of persistence and uncertainty, and income for top earners potentially being highly correlated with the business cycle. In addition, the availability and relevance of public (unemployment) and private (within-family) insurance mechanisms against income fluctuations can vary throughout the income distribution.15 Finally, due to the positive correlation between income and education, top-income individuals may be better aware of current macroeconomic conditions and how they can affect household income. This analysis also allows us to examine the heterogeneity in the welfare consequences of expectation revisions, particularly expectation errors and income uncertainty, which may be more severe for lower-income groups due to limited financial buffers. The three panels in Table 3outline the key estimate results of the bottom-, middleand top-income groups, respectively, with the full set of estimated coefficients shown in Appendix D. First, we detect heterogeneity in the effect of unexpected income changes on the expected value of income (Column 1), possibly reflecting different income processes for the three groups. Approximately 25% of unexpected positive income changes are considered persistent for the bottomand middle-income groups, while top-income 14This measure ensures constant groups and avoids allocating families differently in exceptional years with large unexpected income changes. The average income in the 3 groups is 18,000, 32,000 and 53,000 euros. 15It is worth noting that income refers to net household income. 16 individuals perceive them as transitory (insignificant coefficient in Panel C). In contrast, unexpected negative income changes are significant for all groups but are perceived to be more persistent at the top of the distribution. For top-income earners, a 10% increase in unexpected negative income changes resulting in a 6.5% increase in expected income. We interpret this heterogeneity as arising from differences in labour income dynamics, exposure to the business cycle, and the role of safety nets. A significant fraction of highincome earners, often in managerial roles, come from volatile sources, such as business profits and bonuses, which are tied to business cycle fluctuations. Therefore, they may perceive unexpected negative income changes as more persistent, anticipating a slower recovery after downturns. In addition, their specialized job roles make reemployment more difficult and increase their exposure to business cycle fluctuations. However, Columns 6 and 7 show that this downward revision in expectation is excessive, indicating an overreaction to unexpected income drops in this group. In contrast, they view unexpected income increases as temporary, integrating volatility into their expectations and avoiding expectation errors. For lowand middle-income individuals, unexpected positive and negative income changes have significant and comparable effects on income expectations (Column 1, Panels A and B), and the magnitude is less than half that of the top-income group. Expectation revisions in these groups reflect higher persistence in their income process, which primarily rely on wages and, in some cases, government transfers. Among low-income respondents, unexpected negative income changes have a relatively smaller effect on income expectations, emphasizing the stabilizing role of insurance mechanisms, such as unemployment benefits and transfers, in mitigating the impact of income drops. The determinants of perceived uncertainty (Columns 4-5) also exhibit heterogeneity across the income distribution. First, uncertainty responds to unexpected income changes for lowand middle-income individuals, whereas for top earners, it is primarily driven by aggregate economic conditions. Unexpected positive income changes are the main drivers of perceived income risk for middle-income individuals (Panel B). They revise upward expectations about the future income upper bound but not for the lower bound. This reflects uncertainty about the persistence of the income change. Although they anticipate the possibility of higher future earnings, they do not adjust the lower bound because of concerns about its permanence. As a result, this increases income dispersion and may weaken the consumption response to unexpected positive income changes. At the bottom of the income distribution, perceived uncertainty (Column 4, Panel A) increases only after a negative unexpected income change. However, the magnitude of this effect is roughly one-third of that observed among middle-income individuals, possibly related to a greater role of unemployment benefits and other income support measures among this group. 17 Aggregate conditions significantly affect the dispersion of expectations only among the top-income group, albeit with a relatively modest magnitude.16 The effect of aggregate conditions on top-income earners’ expectations can be related to two main factors. First, high-income earners are often in managerial roles and more exposed to the financial markets, which are more affected by business cycles and macroeconomic fluctuations. This is consistent with the findings of Roth and Wohlfart (2020), suggesting that individuals highly exposed to aggregate risk are more likely to update their personal expectations in response to broader economic conditions. Second, the significant role of aggregate conditions among top-earners may arise from differences in attentiveness and perception of macroeconomic trends, along with the awareness of their impact on household economic conditions. This, in turn, shapes their expectations. This process unfolds in three key stages of expectation formation (Fuster et al.,2022): information selection, information acquisition, and information processing. As shown in Appendix Table D.1, income is positively correlated with education, financial literacy, and the tendency to consult financial sources for decision-making. These factors likely lower the cost of acquiring and processing economic information, making high-income earners more responsive to aggregate conditions.17 Unexpected income changes have different effects on expectation errors across the three subgroups. For the bottomand middle-income groups, the impact of unexpected positive and negative income changes on expectation error is of similar magnitude (Column 6). The reduction in expectation error due to unexpected positive income changes is primarily driven by a weakening of underforecasting, as indicated by the negative coefficients in Column 7. In contrast, unexpected negative income changes lead to both a decrease in overforecasting and an increase in underforecasting.18 Among the top-income group, however, only unexpected negative income changes significantly and substantially increase expectation error. The pronounced downward revision across the entire distribution of income expectations following a negative change suggests an over-reaction in expectations and a shift toward underforecasting in this group, as reflected in the positive coefficients in Columns 6 and 7. In other words, our results suggest that top income respondents underforecast future income, resulting in a large estimated persistence of income changes. This result is in line with the over-persistence bias mechanism in Rozsypal 16The effect of positive unexpected income changes on the standard deviation reported in Column 5 is also statistically significant, albeit with a very small magnitude. 17This finding is consistent with Easaw and Grimme (2024), which highlights that top executives are particularly aware of aggregate uncertainty’s impact on firms, that is likely extends to their household income expectations as well. 18The mixed effect of unexpected negative income changes on over-extrapolation is evident in the reduced estimated effect from Column 6 to Column 7 in Panel A and the statistically insignificant coefficient in Column 7 in Panel B. 18 and Schlafmann (2023) and with empirical findings of over-extrapolation of expectations in Cocco et al. (2022); Massenot and Pettinicchi (2019); D’Acunto and Weber (2024). Unexpected positive changes, on the other hand, are perceived as temporary and do not significantly affect expectation errors. This asymmetry may be attributed to specific features of the income process of top-earners who rely heavily on variable income sources, which are more volatile and correlate with business cycle fluctuations. This distinctive pattern among top-income individuals accounts for the greater average impact of negative income changes compared to positive ones, as shown in Table 2. Overall, our results suggest that after an unexpected positive income change, individuals tend to either not to revise their expectations or to improve their accuracy. However, a significant number of individuals overreact to unexpected negative income changes, excessively revising downward their expectations, particularly at the top of the income distribution. This suggests that the diagnostic expectation mechanism proposed by Bordalo et al. (2018,2019) is especially relevant for high-income individuals. TABLE 3 ABOUT HERE 5. Conclusions We study how unexpected income changes and aggregate conditions affect income expectations, notably their uncertainty and expectation errors. We find that unexpected changes in household income have a significant and relevant impact on expectations revision and their uncertainty, while aggregate conditions play a minor and mostly insignificant role. Results are heterogeneous across the income distribution, possibly due to differences in income processes, the varying relevance of insurance mechanisms, or different levels of awareness regarding how unexpected income changes and aggregate conditions impact household financial conditions. By comparing income expectations and their future realizations we find that revision in expectations is partly due to an over-reaction to unexpected income changes, especially for unexpected negative income changes and among high-income individuals. Our findings help to understand household expectations and, consequently, their behavior in response to unexpected income changes and throughout the business cycle. This, in turn, informs the development of policy interventions, including fiscal and labour market policies. From a welfare perspective, individuals revise their income expectations downward after an unexpected negative income change and upward following an unexpected positive 19 income change, with around 20% of unexpected income changes perceived as persistent. According to the permanent income hypothesis, this induces a change in consumption. If these unexpected income changes are accompanied by an over-reaction of income expectations, consumers make a sub-optimal consumption, which is lower (higher) than its optimum after negative (positive) unexpected income changes. In this line, D’Acunto et al. (2024) report that household spending and debt decisions reflect (inaccurate) subjective income expectations. Our results show that over-reaction to unexpected positive income changes is limited and that the relevance of underforecasts following unexpected negative income changes increases with income. The welfare consequences of suboptimal consumption plans due to expectation errors are less severe for the top income group, characterized by lower marginal utility of consumption and possibly larger buffer stocks. Thus, the ex-ante consumption pattern is closer to the optimal one in the group where consequences of sub-optimality are most pronounced. Prudent individuals also increase their precautionary savings when income uncertainty increases, thereby reducing current consumption. Consumption contraction following an unexpected negative income change is more severe if it is accompanied by an upward revision in income uncertainty. We show evidence of this channel, particularly among low income respondents. This may amplify the consumption contraction after an unexpected income reduction in this group. Unexpected positive income changes are associated with an increase in income dispersion, which weakens the effect of unexpected positive income changes on consumption growth. Evidence of limited responsiveness of household income expectations to aggregate conditions, beyond their individual circumstances, raises concerns about the accurate assessment of future scenarios related to the business cycle. Failure to adequately consider these factors can have detrimental consequences for consumers, particularly in recession periods. Our empirical study has some limitations, which presents opportunities for future research. First, it would be interesting to directly assess how consumption and savings respond to income expectations, uncertainty and expectation error. Unfortunately, the DHS dataset records savings amounts in bands, making this analysis challenging without access to a more detailed dataset. Moreover, we attribute the heterogeneity across the income distribution mainly to differences in the earning process. However, income, education, financial knowledge and portfolio composition are intertwined. Consequently, isolating the specific role of each factor warrants further investigation. Finally, although we observe the correlation between unexpected income changes and expectations, we do not explore the specific channels through which this connection operates. For example, psychological characteristics such as personality traits, or past experiences such as 20 encountering recessions during one’s life cycle, could influence how individuals perceive unexpected income changes. The analysis of underlying mechanisms is left for future research. References Angelini, V. and I. Ferrari (2021). The long-term effects of experienced macroeconomic shocks on wealth. Working Papers 2021:23, Department of Economics, University of Venice "Ca’ Foscari". Armona, L., A. Fuster, and B. Zafar (2019). Home price expectations and behaviour: Evidence from a randomized information experiment. The Review of Economic Studies 86(4 (309)), pp. 1371–1410. Attanasio, O. and B. Augsburg (2016). Subjective expectations and income processes in rural india. Economica 83(331), 416–442. Attanasio, O., A. Kovacs, and K. Molnar (2020). Euler equations, subjective expectations and income shocks. Economica 87(346), 406–441. Bloom, N. (2009). The impact of uncertainty shocks. Econometrica 77(3), 623–685. Bordalo, P., N. Gennaioli, R. La Porta, and A. Shleifer (2019). Diagnostic expectations and stock returns. The Journal of Finance 74(6), 2839–2874. Bordalo, P., N. Gennaioli, and A. Shleifer (2018). Diagnostic expectations and credit cycles. Journal of Finance 73(1), 199–227. Bordalo, P., N. Gennaioli, and A. Shleifer (2022). Overreaction and diagnostic expectations in macroeconomics. Journal of Economic Perspectives 36(3), 223–244. Brown, S., G. Garino, and K. Taylor (2008). Mortgages and financial expectations: A household-level analysis. Southern Economic Journal 74(3), 857–878. Brown, S. and K. Taylor (2006). Financial expectations, consumption and saving: A microeconomic analysis. Fiscal Studies 27(3), 313–338. Bucciol, A. and R. Miniaci (2018). Financial risk propensity, business cycles and perceived risk exposure. Oxford Bulletin of Economics and Statistics 80(1), 160–183. Bucciol, A. and L. Zarri (2015). The shadow of the past: Financial risk taking and negative life events. Journal of Economic Psychology 48, 1–16. 21 Christelis, D., D. Georgarakos, T. Jappelli, and M. van Rooij (2020). Consumption uncertainty and precautionary saving. The Review of Economics and Statistics 102(1), 148–161. Cocco, J., F. Gomes, and P. Lopes-Cocco (2022). Evidence on expectations of household finances. CEPR discussion paper 17447, Center for Economic Policy Research. Coibion, O., D. Georgarakos, Y. Gorodnichenko, G. Kenny, and M. Weber (2024). The effect of macroeconomic uncertainty on household spending. American Economic Review 114(3), 645–77. Coibion, O., Y. Gorodnichenko, and M. Weber (2021). Fiscal policy and households’ inflation expectations: Evidence from a randomized control trial. NBER working paper 28485, National Bureau of Economic Research. D’Acunto, F. and M. Weber (2024). Why survey-based subjective expectations are meaningful and important. Annual Review of Economics 16. Dominitz, J. and C. F. Manski (1997). Using expectations data to study subjective income expectations. Journal of the American Statistical Association 92(439), 855–867. Dominitz, J. and C. F. Manski (2004). How should we measure consumer confidence? Journal of Economic Perspectives 18(2), 51–66. D’Acunto, F., M. Weber, and X. Yin (2024, July). Subjective income expectations and household debt cycles. Working Paper 32715, National Bureau of Economic Research. Easaw, J. and C. Grimme (2024). The relationship between aggregate uncertainty and firm-level uncertainty. Oxford Economic Papers gpae004, 1108–1127. Enke, B. and T. Graeber (2023, 05). Cognitive uncertainty. The Quarterly Journal of Economics 138(4), 2021–2067. Fagereng, A., L. Guiso, and L. Pistaferri (2018). Portfolio choices, firm shocks, and uninsurable wage risk. The Review of Economic Studies 85(1), 437–474. Fuster, A., R. Perez-Truglia, M. Wiederholt, and B. Zafar (2022). Expectations with endogenous information acquisition: An experimental investigation. The Review of Economics and Statistics 104(5), 1059–1078. Gennaioli, N. and A. Shleifer (2010). What comes to mind. The Quarterly Journal of Economics 125(4), 1399–1433. 22 Figure 3: Time pattern of unemployment rate and macroeconomic uncertainty Notes: The graph shows the (3-months average) unemployment rate and the Policy Uncertainty Index (monthly values, ihs). For the latter, it plots both the original data points (dotted line) and those obtained by applying a smoothness filter (local OLS regression implemented through the lowess command in Stata; solid line). 29 Figure 4: Marginal effects of small and large unexpected income changes Notes: Estimated coefficients and 95% standard errors. Complete estimate results are reported in Appendix Table C.5. 30 A. Appendix: Variable definition and sensitivity checks A.1. Income expectations We derive income expectation (variable Exp. y in the analysis) as a weighted average using the probabilities PRO1,PRO2,PRO3 and PRO4 and the associated amounts. We otherwise take the simple average between LAAG and HOOG in case LAAG and HOOG differ by less than 5 euros. We also focus on the lower and upper bounds of income expectation as an outcome of the analysis. They are, respectively, variables LB and UB in the analysis. To further explore the relationship between income expectations and job-related expectations, we use additional information collected by the DHS. Respondents, categorized according to their employment status, are asked about the probability of losing or finding a job in the next 12 months. We estimate conditional correlations through OLS regressions of income on the probability of job loss or job finding while controlling for working status and a set of covariates. Results for working and unemployed individuals are graphically summarized in Figure A.1. The perceived probability of job loss is significantly correlated with most outcome variables, displaying the expected sign. The results for the unemployed subgroup are less precise, partly due to the smaller sample size. However, the upper bound of expected income and income uncertainty are significantly correlated with the likelihood of finding a job. These findings support the primary role of labour income in shaping total household income expectations. 31 Figure A.1: Correlation between outcome variables and job-related expectations Notes: Conditional correlation between outcome variables and job-related expectations. The graph plots OLS estimated coefficients and 90% level confidence intervals. The dependent variables are the same as in Table 2, and the key independent variable is the probability of losing/finding a job for workers or unemployed, respectively. Control variables are the same as in Table 2. A.2. Expectation uncertainty We consider two main measures for income uncertainty. The first is the difference between the upper and lower bounds of the income expectations (variable UB-LB in the analysis). We also create a measure of standard deviation by exploiting the nature of the data. The standard deviation of expected income (variable SD exp. in the analysis) is derived from the probabilities and the associated amounts in questions PRO1-PRO4. The standard deviation is otherwise set to zero if LAAG and HOOG differ by less than 5 euros. A.3. Expectation error We define the expectation error (variable Exp. err. in the analysis) as the difference between the income realization reported in year t+1 and the income expectation for year t+ 1 reported in year t. We also consider its absolute value (variable Exp. err. (abs)) to focus on the magnitude of the expectation error. The baseline sample includes respondents who give “consistent” responses on the probability distribution of expected income, namely those who are either i) certain about their future income (the difference between upper and lower bounds is smaller than 5 euros) or ii) reporting increasing probabilities with expected income thresholds. Hence, 83.18% of the respondents give consistent probabilities (or are certain about future income). Even if 32 less than 17% of the respondents report inconsistent probabilities, this may raise concerns about sample selection. To address this issue, we first examine the factors associated with the probability of giving a consistent probability distribution. OLS regression results are reported in Table A.1. We only find a significant correlation with gender and age. Table A.1: Sample selection: Probability of giving a consistent probability distribution (1) Dep. var. Consistent answer Age 0.002** (0.001) Partner in the hh -0.006 (0.015) Children in the hh 0.011 (0.016) Working 0.006 (0.020) Retired 0.012 (0.020) Homeowner 0.010 (0.014) Female 0.053*** (0.016) Primary -0.008 (0.036) High school 0.026 (0.035) Vocational training 0.016 (0.037) University 0.032 (0.037) Income realization -0.006 (0.006) Financial assets -0.000 (0.000) Year FE Yes Constant 0.760*** (0.083) Observations 4,620 R-squared 0.021 Notes: Standard errors in parentheses. ∗∗∗p<0.01,∗ ∗ p<0.05,∗p<0.1. Second, we select the outcome variables that are not affected by reported probabilities (lower bound, upper bound and their difference), and we run the same regressions shown in Table 2. Results reported in Table A.2 are consistent with the benchmark results. A.4. Sensitivity analysis We assess robustness of results in Table 2in two alternative samples. Table A.3 reports estimate results for the sample that includes partners in addition to heads. Table A.4 also incorporates respondents reporting income bands for household income in addition 33 Table A.2: Sample including respondents with inconsistent probabilities (comparable outcomes) (1) (2) (3) Dep. var. LB UB UB-LB Unexp. positive ∆y 0.129*** 0.185*** 0.056*** (0.021) (0.018) (0.015) Unexp. negative ∆y (abs) -0.265*** -0.221*** 0.045*** (0.024) (0.020) (0.016) Unemployed 0.054 -0.157 -0.211* (0.185) (0.153) (0.126) Uncertainty in NL 0.059 0.042 -0.018 (0.057) (0.048) (0.039) Unempl. rate -0.016 -0.020 -0.004 (0.016) (0.013) (0.011) Age 0.030** 0.027** -0.003 (0.013) (0.011) (0.009) Partner in the hh 0.147 0.100 -0.048 (0.157) (0.130) (0.107) Children in the hh 0.130 -0.027 -0.157* (0.118) (0.098) (0.080) Working 0.266* 0.083 -0.183* (0.151) (0.125) (0.103) Retired 0.126 -0.064 -0.190* (0.143) (0.119) (0.098) Homeowner 0.158 0.197 0.039 (0.201) (0.167) (0.137) Constant 8.380*** 9.071*** 0.691 (0.996) (0.828) (0.679) R-squared 0.055 0.078 0.008 Number of individuals 1,190 1,190 1,190 Observations 4,620 4,620 4,620 Notes: Standard errors in parentheses. ∗∗∗p<0.01,∗ ∗ p<0.05,∗p<0.1. 34 to respondents reporting precise income values.A.1 Our key results are confirmed in both alternative samples. Table A.3: Sample including partners (1) (2) (3) (4) (5) (6) (7) Dep. Var Exp. y LB UB UB-LB SD exp. Exp. err. Exp. err. (abs) Unexp. positive ∆y 0.157*** 0.126*** 0.161*** 0.035** 0.004*** -0.120*** -0.114*** (0.017) (0.021) (0.018) (0.015) (0.001) (0.025) (0.022) Unexp. negative ∆y (abs) -0.212*** -0.250*** -0.208*** 0.043*** 0.004*** 0.335*** -0.002 (0.018) (0.022) (0.018) (0.015) (0.001) (0.026) (0.023) Unemployed -0.124 0.115 -0.145 -0.260** -0.011 0.308 -0.044 (0.151) (0.186) (0.153) (0.128) (0.009) (0.221) (0.195) Uncertainty in NL 0.014 0.022 0.008 -0.015 0.002 -0.069 -0.036 (0.046) (0.057) (0.047) (0.039) (0.003) (0.068) (0.060) Unempl. rate -0.028** -0.020 -0.020 -0.001 0.003*** 0.028 0.020 (0.012) (0.015) (0.013) (0.011) (0.001) (0.018) (0.016) Age 0.021** 0.020 0.018* -0.002 -0.000 -0.013 -0.018 (0.010) (0.013) (0.011) (0.009) (0.001) (0.015) (0.014) Partner in the hh 0.185 0.443*** 0.175 -0.267** -0.012 0.156 -0.059 (0.138) (0.170) (0.140) (0.117) (0.008) (0.202) (0.179) Children in the hh 0.078 0.142 0.062 -0.080 -0.009* -0.005 0.165 (0.096) (0.118) (0.097) (0.081) (0.005) (0.140) (0.124) Working 0.139 0.070 0.134 0.063 0.005 0.296* -0.201 (0.113) (0.139) (0.115) (0.096) (0.006) (0.165) (0.146) Retired -0.050 -0.050 -0.065 -0.015 -0.006 0.345** -0.093 (0.110) (0.136) (0.112) (0.094) (0.006) (0.161) (0.142) Homeowner 0.370** 0.341* 0.360** 0.019 -0.008 -0.395* -0.129 (0.153) (0.188) (0.155) (0.130) (0.009) (0.224) (0.198) Constant 9.308*** 8.970*** 9.531*** 0.561 0.045 0.790 1.905* (0.810) (0.999) (0.823) (0.689) (0.046) (1.185) (1.048) R-squared 0.074 0.054 0.070 0.008 0.019 0.058 0.010 Number of individuals 1,447 1,447 1,447 1,447 1,447 1,447 1,447 Observations 4,917 4,917 4,917 4,917 4,917 4,917 4,917 Notes: Standard errors in parentheses. ∗∗∗p<0.01,∗ ∗ p<0.05,∗p<0.1. A.1In particular, we rely on the answer to question: “Please indicate about how much the total net income of your household was over the period 1 January [year] through 31 December [year].” In this case, possible answers are a set of thresholds ranging from 1 (less than 8,000 euros) to 11 (more than 75,000 euros). For instance, threshold 5 indicates incomes between 13,000 and 16,000 euros. We use for observed income the intermediate threshold value; extreme thresholds are set at their boundaries (i.e. 8,000 euros for threshold 1 and 75,000 euros for threshold 11). 35 Table A.4: Sample including income in brackets (1) (2) (3) (4) (5) (6) (7) Dep. var. Exp. y LB UB UB-LB SD exp. Exp. err. Exp. err. (abs) Unexp. positive ∆y 0.112*** 0.103*** 0.112*** 0.009 0.001 -0.074*** -0.087*** (0.014) (0.016) (0.015) (0.010) (0.001) (0.019) (0.016) Unexp. negative ∆y (abs) -0.202*** -0.219*** -0.204*** 0.015 -0.002 0.340*** 0.027 (0.023) (0.027) (0.024) (0.017) (0.002) (0.029) (0.025) Unemployed 0.167 0.069 0.202 0.132 0.037** 0.459* 0.044 (0.204) (0.233) (0.207) (0.147) (0.016) (0.250) (0.218) Uncertainty in NL -0.047 -0.031 -0.050 -0.019 0.002 0.009 -0.061 (0.057) (0.066) (0.058) (0.041) (0.005) (0.072) (0.062) Unempl. rate -0.043*** -0.029 -0.039** -0.010 0.000 0.044** 0.010 (0.016) (0.018) (0.016) (0.012) (0.001) (0.020) (0.017) Age 0.005 0.010 0.003 -0.008 -0.001 0.003 -0.008 (0.013) (0.015) (0.013) (0.009) (0.001) (0.016) (0.014) Partner in the hh 0.042 0.019 0.023 0.004 0.002 0.135 -0.085 (0.159) (0.182) (0.161) (0.115) (0.012) (0.196) (0.171) Children in the hh 0.144 0.357*** 0.111 -0.245*** -0.043*** 0.008 0.166 (0.118) (0.135) (0.120) (0.085) (0.009) (0.145) (0.127) Working 0.290* 0.344* 0.277* -0.067 -0.004 0.272 -0.247 (0.164) (0.188) (0.167) (0.119) (0.013) (0.198) (0.173) Retired 0.153 0.202 0.143 -0.059 -0.005 0.219 -0.270 (0.161) (0.184) (0.163) (0.116) (0.013) (0.193) (0.168) Homeowner -0.313* 0.284 -0.364** -0.648*** -0.095*** 0.078 0.290 (0.182) (0.208) (0.185) (0.132) (0.014) (0.229) (0.199) Constant 10.706*** 9.515*** 10.985*** 1.471** 0.158** -0.692 1.440 (0.972) (1.110) (0.986) (0.702) (0.076) (1.225) (1.066) R-squared 0.031 0.025 0.031 0.008 0.016 0.036 0.009 Number of individuals 2,114 2,114 2,114 2,114 2,114 1,779 1,779 Observations 7,637 7,637 7,637 7,637 7,637 6,527 6,527 Notes: Standard errors in parentheses. ∗∗∗p<0.01,∗ ∗ p<0.05,∗p<0.1. 36 B. Appendix: Omitted variable test We use the method developed by Oster (2019) to evaluate the possible degree of omitted variable bias under the assumption that the selection on the observed controls is correlated with the selection of the observables. The method in Oster (2019) allows us to address selection bias for one critical variable only. For this reason, we do not distinguish between positive and negative unexpected income changes, but we include a single regressor for the inverse hyperbolic sine of the unexpected income change.B.1 Results are reported in Table B.1. Following the parametrization suggested by Oster (2019), we assume that the degree of variation which both observed and unobserved variables can account for is proportional to the variance explained by the covariates.B.2 The bottom line in Table B.1 reports the degree of selection on unobservables relative to observables (the parameter δ) that would be necessary to explain away the results. The absolute value of δalways exceeds the rule of thumb cut-off of 1 indicated by Oster (2019). These findings strongly support the robustness of our findings to the omitted variable bias. B.1We also include, alternatively, the positive and negative unexpected income changes. The main findings are confirmed. B.2More precisely, we assume that Rmax = 1.3˜ R, where Rmax is the R2obtained in the hypothetical regression of the dependent variable on both observed and unobserved regressors; ˜ Ris the R2of the regression of the dependent variable on observables. 37 Table B.1: Oster test on omitted variable bias (1) (2) (3) (4) (5) (6) (7) Dep. var. Exp. y LB UB UB-LB SD exp. Exp. err. Exp. err. (abs) Unexp. ∆y 0.222*** 0.214*** 0.225*** 0.011 0.002** -0.270*** -0.082*** (0.015) (0.018) (0.015) (0.012) (0.001) (0.020) (0.018) Uncertainty in NL 0.031 0.048 0.026 -0.022 0.001 -0.067 -0.064 (0.057) (0.068) (0.057) (0.046) (0.003) (0.078) (0.069) Unempl. rate -0.024 -0.006 -0.017 -0.012 0.002** 0.031 0.011 (0.015) (0.018) (0.015) (0.012) (0.001) (0.021) (0.018) Age 0.027** 0.027* 0.024* -0.003 -0.001 -0.025 -0.016 (0.013) (0.015) (0.013) (0.010) (0.001) (0.017) (0.015) Partner in the hh 0.084 0.105 0.080 -0.025 -0.007 0.226 -0.032 (0.159) (0.190) (0.161) (0.129) (0.008) (0.220) (0.194) Children in the hh -0.002 0.195 -0.027 -0.222** -0.017*** -0.055 0.234* (0.114) (0.137) (0.116) (0.092) (0.006) (0.158) (0.139) Working 0.219** 0.245* 0.207* -0.037 -0.003 0.082 -0.199 (0.111) (0.133) (0.113) (0.090) (0.006) (0.154) (0.136) Retired -0.013 0.035 -0.029 -0.064 -0.008 0.188 -0.045 (0.117) (0.140) (0.119) (0.095) (0.006) (0.162) (0.143) Homeowner 0.224 0.178 0.226 0.048 -0.001 -0.223 -0.140 (0.192) (0.230) (0.195) (0.156) (0.010) (0.267) (0.235) Constant 8.905*** 8.539*** 9.148*** 0.609 0.077 1.523 1.913 (0.983) (1.177) (0.997) (0.794) (0.051) (1.362) (1.201) Oster delta 89.241 -59.883 44.5152 37.177 33.654 -15.262 19.125 R-squared 0.083 0.055 0.082 0.004 0.012 0.062 0.010 Number of individuals 1,064 1,064 1,064 1,064 1,064 1,064 1,064 Observations 3,767 3,767 3,767 3,767 3,767 3,767 3,767 Notes: Standard errors in parentheses. ∗∗∗p<0.01,∗ ∗ p<0.05,∗p<0.1. 38 Table C.5: Benchmark analysis using unexpected income change size (1) (2) (3) (4) (5) (6) (7) VARIABLES Exp. y LB UB UB-LB SD exp For. err. For. err. (abs) If unexp. pos. ∆y -0.032 0.036 -0.044 -0.080 -0.005 -0.082 0.226** (0.083) (0.099) (0.084) (0.066) (0.004) (0.114) (0.101) If large pos. ∆y 0.006 0.057 0.008 -0.048 0.002 -0.074 -0.019 (0.072) (0.085) (0.073) (0.058) (0.004) (0.099) (0.087) If large neg. ∆y -0.015 0.138 -0.019 -0.157** -0.002 -0.101 0.052 (0.076) (0.091) (0.078) (0.061) (0.004) (0.105) (0.093) Unexp. positive ∆y -0.000 0.160 0.047 -0.114 0.000 0.132 -0.825 (0.851) (1.016) (0.864) (0.685) (0.044) (1.174) (1.037) Unexp. negative ∆y (abs) -0.721 -0.552 -0.771 -0.219 -0.009 -1.182 2.586** (0.910) (1.086) (0.924) (0.732) (0.047) (1.255) (1.109) Unexp. positive ∆y*If large pos. ∆y 0.201 -0.027 0.161 0.188 0.004 -0.280 0.688 (0.852) (1.017) (0.865) (0.685) (0.044) (1.174) (1.038) Unexp. negative ∆y (abs)*If large neg. ∆y 0.468 0.229 0.518 0.289 0.010 1.577 -2.513** (0.910) (1.087) (0.924) (0.732) (0.047) (1.255) (1.110) Unemployed -0.154 0.018 -0.166 -0.184 -0.002 0.360 -0.077 (0.182) (0.217) (0.185) (0.146) (0.009) (0.251) (0.222) Uncertainty in NL 0.032 0.046 0.026 -0.020 0.001 -0.067 -0.061 (0.057) (0.068) (0.057) (0.046) (0.003) (0.078) (0.069) Unempl. rate -0.024 -0.007 -0.017 -0.010 0.002** 0.031 0.012 (0.015) (0.018) (0.015) (0.012) (0.001) (0.021) (0.018) Age 0.027** 0.026* 0.024* -0.002 -0.001 -0.022 -0.017 (0.013) (0.015) (0.013) (0.010) (0.001) (0.017) (0.015) Partner in the hh 0.094 0.155 0.089 -0.067 -0.008 0.199 -0.045 (0.160) (0.191) (0.162) (0.128) (0.008) (0.220) (0.195) Children in the hh -0.006 0.185 -0.031 -0.216** -0.017*** -0.044 0.231* (0.114) (0.136) (0.116) (0.092) (0.006) (0.157) (0.139) Working 0.140 0.264 0.121 -0.143 -0.004 0.250 -0.221 (0.145) (0.173) (0.147) (0.117) (0.007) (0.200) (0.177) Retired -0.070 0.074 -0.093 -0.167 -0.010 0.296 -0.055 (0.141) (0.168) (0.143) (0.113) (0.007) (0.194) (0.172) Homeowner 0.228 0.192 0.229 0.036 -0.002 -0.263 -0.105 (0.193) (0.230) (0.196) (0.155) (0.010) (0.266) (0.235) Constant 9.045*** 8.555*** 9.302*** 0.746 0.080 1.298 1.812 (0.991) (1.183) (1.006) (0.797) (0.051) (1.366) (1.207) R-squared 0.084 0.062 0.083 0.015 0.016 0.073 0.016 Number of individuals 1,064 1,064 1,064 1,064 1,064 1,064 1,064 Observations 3,767 3,767 3,767 3,767 3,767 3,767 3,767 Notes: Standard errors in parentheses. ∗ ∗ ∗p<0.01,∗ ∗ p<0.05,∗p<0.1.Large positive and negative unexpected income changes are defined as unexpected income change larger than their respective median. 45 Table C.6: Benchmark analysis including lagged unexpected income changes (1) (2) (3) (4) (5) (6) (7) Dep. var Exp. y LB UB UB-LB SD exp. Exp. err. Exp. err. (abs) Unexp. positive ∆y 0.211*** 0.117*** 0.215*** 0.098*** 0.004** -0.205*** -0.140*** (0.030) (0.037) (0.031) (0.026) (0.002) (0.042) (0.037) Unexp. negative ∆y (abs) -0.356*** -0.383*** -0.351*** 0.031 0.003* 0.505*** 0.144*** (0.028) (0.034) (0.029) (0.024) (0.002) (0.039) (0.034) Unexp. positive ∆y (lag) 0.094*** 0.133*** 0.094*** -0.039 -0.001 -0.017 -0.167*** (0.027) (0.033) (0.028) (0.024) (0.002) (0.038) (0.033) Unexp. negative ∆y (abs, lag) -0.153*** -0.118*** -0.154*** -0.036 -0.001 0.259*** 0.041 (0.032) (0.039) (0.032) (0.027) (0.002) (0.044) (0.038) Unemployed -0.155 -0.181 -0.130 0.051 0.012 0.565* -0.278 (0.221) (0.270) (0.226) (0.193) (0.013) (0.309) (0.269) Uncertainty in NL -0.085 -0.082 -0.093 -0.011 -0.001 0.092 -0.001 (0.071) (0.086) (0.072) (0.062) (0.004) (0.099) (0.086) Unempl. rate 0.031 0.035 0.036* 0.000 0.001 -0.023 -0.012 (0.021) (0.025) (0.021) (0.018) (0.001) (0.029) (0.025) Age -0.002 -0.000 -0.005 -0.005 -0.001 0.013 -0.004 (0.016) (0.020) (0.016) (0.014) (0.001) (0.022) (0.019) Partner in the hh 0.161 0.122 0.183 0.061 0.011 -0.030 -0.349 (0.234) (0.285) (0.239) (0.203) (0.014) (0.327) (0.284) Children in the hh 0.055 0.147 0.009 -0.138 -0.020* -0.081 0.028 (0.175) (0.214) (0.179) (0.153) (0.011) (0.245) (0.213) Working -0.096 -0.133 -0.075 0.058 0.018 0.524** -0.090 (0.189) (0.231) (0.193) (0.165) (0.012) (0.265) (0.230) Retired 0.047 0.046 0.048 0.002 0.002 0.240 -0.218 (0.177) (0.216) (0.181) (0.154) (0.011) (0.247) (0.215) Homeowner 0.108 0.085 0.114 0.029 0.005 -0.122 -0.022 (0.256) (0.313) (0.262) (0.223) (0.016) (0.358) (0.311) Constant 11.171*** 10.958*** 11.402*** 0.444 0.072 -1.520 1.233 (1.247) (1.525) (1.275) (1.087) (0.076) (1.744) (1.517) R-squared 0.162 0.112 0.155 0.019 0.020 0.148 0.042 Number of individuals 706 706 706 706 706 706 706 Observations 2,032 2,032 2,032 2,032 2,032 2,032 2,032 Notes: Standard errors in parentheses. ∗∗∗p<0.01,∗ ∗ p<0.05,∗p<0.1. 46 D. Appendix: Additional tables on heterogeneity This Appendix presents the full set of estimated coefficients corresponding to the results in Table 3, along with descriptive statistics of the three subsamples. More precisely, we report the characteristics of the bottom and top 33% income groups (see Table D.1) and the benchmark analysis split by sample group: Bottom 33% (see Table D.2), middle 33% (see Table D.3) and top 33% (see Table D.4). Table D.1: Characteristics in the bottom and top 33% income groups Variable Label Bottom Top t-test Income variables Expected income Exp. y 10.301 11.364 -22.098*** Lower bound exp. inc. LB 10.139 11.222 -19.365 *** Upper bound exp. inc. UB 10.355 11.420 -21.910*** Upper - Lower bound UB-LB 0.216 0.198 0.524 SD expected income SD exp. 0.028 0.034 -2.826*** Expectation error Exp. err. -0.130 0.045 -3.163*** Expectation error (abs.) Exp. err. (abs) 0.716 0.360 7.018*** Key explanatory variables Unexpected positive income change Unexp. positive ∆y 0.178 0.169 0.571 Unexpected negative income change (abs.) Unexp. negative ∆y (abs) 0.283 0.103 9.854*** Unemployed 0.039 0.005 5.864*** Uncertainty in NL 5.002 4.973 1.200 Unempl. rate 5.570 5.639 -1.347 Control variables Age 61.545 58.073 7.136*** Partner in the hh 0.449 0.885 -26.315*** Children in the hh 0.149 0.275 -7.792*** Working 0.298 0.576 -14.537*** Retired 0.434 0.396 1.922* Homeowner 0.567 0.931 -23.376*** Further variables Female 0.360 0.116 15.078*** College educ. 0.052 0.301 -17.030*** Vocational training educ. 0.219 0.097 8.491*** High School educ. 0.323 0.470 -7.568*** Low educ. 0.364 0.112 15.620*** No educ. 0.035 0.013 3.638*** Financial literate 0.287 0.527 -12.456*** Media financial source 0.423 0.589 -8.336*** Income (thousands) 18.273 52.181 -43.111*** Financial assets (thousands) 38.236 93.518 -10.605*** Observations 1,197 1,304 Notes: The last column reports the value of a t-test comparing the mean of the bottom and top 33% of the income distribution. ∗∗∗p<0.01,∗ ∗ p<0.05,∗p<0.1. 47 Table D.2: Subsample of bottom 33% income earners (full output) (1) (2) (3) (4) (5) (6) (7) Dep. var. Exp. y LB UB UB-LB SD exp. Exp. err. Exp. err. (abs) Unexp. positive ∆y 0.265*** 0.241*** 0.269*** 0.028 0.002 -0.201*** -0.110** (0.038) (0.044) (0.039) (0.030) (0.002) (0.054) (0.045) Unexp. negative ∆y (abs) -0.098** -0.173*** -0.099** 0.074** 0.001 0.263*** -0.129*** (0.038) (0.044) (0.039) (0.030) (0.002) (0.054) (0.045) Unemployed -0.196 0.024 -0.281 -0.304 -0.033* 0.547 -0.216 (0.342) (0.397) (0.348) (0.264) (0.017) (0.480) (0.405) Uncertainty in NL -0.035 0.019 -0.047 -0.066 -0.003 -0.149 -0.082 (0.134) (0.155) (0.136) (0.103) (0.007) (0.188) (0.158) Unempl. rate -0.043 -0.026 -0.038 -0.012 0.000 0.044 0.011 (0.035) (0.041) (0.036) (0.027) (0.002) (0.049) (0.042) Age 0.040 0.047 0.032 -0.014 -0.002 -0.020 -0.064* (0.030) (0.035) (0.031) (0.023) (0.002) (0.042) (0.036) Partner in the hh -0.428 -0.251 -0.466 -0.214 -0.050** 0.740 -0.352 (0.391) (0.454) (0.398) (0.303) (0.020) (0.549) (0.464) Children in the hh -0.219 -0.016 -0.293 -0.277 -0.051*** 0.096 0.469 (0.312) (0.362) (0.317) (0.241) (0.016) (0.438) (0.370) Working 0.547* 0.813** 0.494* -0.319 -0.018 -0.154 -0.763** (0.282) (0.328) (0.287) (0.218) (0.014) (0.396) (0.335) Retired -0.173 0.039 -0.184 -0.223 -0.011 0.310 0.062 (0.262) (0.304) (0.267) (0.203) (0.013) (0.368) (0.311) Homeowner 0.301 0.100 0.294 0.195 0.004 -0.702 0.097 (0.602) (0.699) (0.613) (0.466) (0.031) (0.846) (0.714) Constant 8.189*** 7.125** 8.795*** 1.670 0.203* 1.502 5.330* (2.379) (2.763) (2.422) (1.840) (0.121) (3.342) (2.824) R-squared 0.096 0.079 0.093 0.015 0.028 0.063 0.031 Number of individuals 390 390 390 390 390 390 390 Observations 1,197 1,197 1,197 1,197 1,197 1,197 1,197 Notes: Standard errors in parentheses. The sample includes respondents with average income in the bottom 33% of the distribution. ∗∗∗p<0.01,∗ ∗ p<0.05,∗p<0.1. 48 Table D.3: Subsample of middle 33% income earners (full output) (1) (2) (3) (4) (5) (6) (7) Dep. var. Exp. y LB UB UB-LB SD exp. Exp. err. Exp. err. (abs) Unexp. positive ∆y 0.245*** 0.071 0.254*** 0.183*** 0.010*** -0.265*** -0.220*** (0.034) (0.043) (0.034) (0.033) (0.002) (0.050) (0.045) Unexp. negative ∆y (abs) -0.239*** -0.263*** -0.236*** 0.028 0.003 0.371*** 0.099 (0.045) (0.059) (0.047) (0.045) (0.003) (0.067) (0.060) Unemployed -0.045 0.062 0.020 -0.042 0.028* -0.158 0.135 (0.235) (0.303) (0.241) (0.231) (0.015) (0.348) (0.312) Uncertainty in NL 0.082 0.064 0.066 0.002 -0.007 -0.059 -0.076 (0.077) (0.099) (0.078) (0.075) (0.005) (0.113) (0.102) Unempl. rate -0.033 -0.005 -0.027 -0.022 0.002 0.027 0.041 (0.021) (0.027) (0.021) (0.020) (0.001) (0.030) (0.027) Age 0.029* 0.030 0.027 -0.003 -0.002* -0.028 -0.011 (0.017) (0.022) (0.017) (0.017) (0.001) (0.025) (0.022) Partner in the hh 0.282 0.279 0.271 -0.008 0.003 -0.112 0.033 (0.191) (0.246) (0.196) (0.188) (0.012) (0.283) (0.254) Children in the hh 0.056 0.429** 0.048 -0.382** -0.009 -0.063 0.257 (0.158) (0.204) (0.162) (0.156) (0.010) (0.234) (0.210) Working -0.057 -0.075 -0.025 0.050 0.008 0.063 0.295 (0.198) (0.255) (0.203) (0.195) (0.012) (0.293) (0.263) Retired -0.123 -0.092 -0.119 -0.028 0.003 0.071 0.311 (0.196) (0.252) (0.201) (0.193) (0.012) (0.290) (0.260) Homeowner 0.228 0.287 0.212 -0.075 -0.021 -0.007 -0.202 (0.212) (0.273) (0.217) (0.208) (0.013) (0.314) (0.281) Constant 8.681*** 8.306*** 8.905*** 0.600 0.168** 1.826 1.085 (1.333) (1.717) (1.365) (1.311) (0.084) (1.975) (1.771) R-squared 0.102 0.040 0.099 0.041 0.040 0.074 0.037 Number of individuals 343 343 343 343 343 343 343 Observations 1,266 1,266 1,266 1,266 1,266 1,266 1,266 Notes: Standard errors in parentheses. The sample includes respondents with average income in the middle 33% of the distribution. ∗∗∗p<0.01,∗ ∗ p<0.05,∗p<0.1. 49 Table D.4: Subsample of top 33% income earners (full output) (1) (2) (3) (4) (5) (6) (7) Dep. var. Exp. y LB UB UB-LB SD exp. Exp. err. Exp. err. (abs) Unexp. positive ∆y 0.010 0.005 0.015 0.010 0.004** 0.030 -0.010 (0.039) (0.047) (0.039) (0.029) (0.002) (0.050) (0.047) Unexp. negative ∆y (abs) -0.649*** -0.632*** -0.650*** -0.017 -0.000 0.787*** 0.423*** (0.044) (0.053) (0.044) (0.033) (0.002) (0.057) (0.054) Unemployed -0.552 -0.533 -0.555 -0.022 0.003 1.269** -0.067 (0.432) (0.526) (0.434) (0.327) (0.019) (0.562) (0.529) Uncertainty in NL 0.007 0.019 0.014 -0.006 0.009*** 0.017 -0.016 (0.081) (0.099) (0.081) (0.061) (0.004) (0.106) (0.099) Unempl. rate -0.003 0.003 0.004 0.002 0.002** 0.022 -0.007 (0.022) (0.026) (0.022) (0.016) (0.001) (0.028) (0.027) Age 0.013 0.001 0.012 0.011 0.001 -0.015 0.007 (0.018) (0.022) (0.018) (0.014) (0.001) (0.024) (0.023) Partner in the hh 0.358 0.287 0.389 0.102 0.025** 0.082 0.051 (0.266) (0.324) (0.267) (0.201) (0.012) (0.346) (0.326) Children in the hh 0.043 0.147 0.025 -0.122 -0.008 -0.117 0.132 (0.148) (0.180) (0.149) (0.112) (0.006) (0.193) (0.182) Working -0.412 -0.400 -0.460 -0.060 -0.001 1.380*** -0.160 (0.312) (0.379) (0.313) (0.236) (0.014) (0.406) (0.382) Retired -0.448 -0.360 -0.500 -0.140 -0.011 1.180*** -0.166 (0.307) (0.374) (0.308) (0.232) (0.013) (0.399) (0.376) Homeowner 0.019 0.006 0.051 0.046 0.023* -0.166 -0.212 (0.309) (0.377) (0.311) (0.234) (0.014) (0.402) (0.379) Constant 10.732*** 11.217*** 10.759*** -0.458 -0.116* -0.589 0.310 (1.438) (1.750) (1.443) (1.087) (0.063) (1.870) (1.761) R-squared 0.191 0.132 0.191 0.006 0.051 0.172 0.066 Number of individuals 331 331 331 331 331 331 331 Observations 1,304 1,304 1,304 1,304 1,304 1,304 1,304 Notes: Standard errors in parentheses. The sample includes respondents with average income in the top 33% of the distribution. ∗∗∗p<0.01,∗ ∗ p<0.05,∗p<0.1. 50 MUNI Econ Working Paper Series (since 2018) 2025-04 Bucciol, A., Easaw, J., Trucchi, S. (2025). Household Income Expectations: The Role of Unexpected Income Changes and Aggregate Conditions. MUNI ECON Working Paper n. 2025-04. Brno: Masaryk University. https://doi.org/10.5817/WP_MUNI_ECON_2025-04 2025-03 Sarsenbayeva, A.; Alpysbayeva, D. (2025). Catastrophic Health Expenditure during Healthcare Financing Reform: Evidence from Kazakhstan. https://doi.org/10.5817/WP_MUN_ECON_2025-03 2025-02 Scervini, F., Trucchi, S. (2025). Alcohol Consumption in an Empty Nest. MUNI ECON Working Paper n. 2025-02. Brno: Masaryk University. https://doi.org/10.5817/WP_MUN_ECON_2025-02 2025-01 Soucek, C., Reggiani, T., Kairies-Schwarz, N. (2025). Physicians’ Responses to Time Pressure: Experimental Evidence on Treatment Quality and Documentation Behaviour. MUNI ECON Working Paper n. 2025-01. Brno: Masaryk University. https://doi.org/10.5817/WP_MUNI_ECON_2025-01 2024-06 Fazio, A., Reggiani, T., Santori, P. (2024). ‘Blessed are the Poor’ The Weberian Spirit of Capitalism Under Experimental Scrutiny. MUNI ECON Working Paper n. 2024-06. Brno: Masaryk University. https://doi.org/10.5817/WP_MUNI_ECON_2024-06 2024-05 Pulina, M., Salis, A. (2024). Dynamics and trends of drug dealing: a local labour system perspective. MUNI ECON Working Paper n. 2024-05. Brno: Masaryk University. https://doi.org/10.5817/WP_MUNI_ECON_2024-05 2024-04 Coufalová, L., Kecskésová, M., Mikula, Š., Ševčík, M. (2024). Does Democracy Flourish in the Dark? Regional Development and Democracy Building. MUNI ECON Working Paper n. 2024-04. Brno: Masaryk University. https://doi.org/10.5817/WP_MUNI_ECON_2024-04 2024-03 Coufalová, L., Dellinger, F. H., Huber, P., Mikula, Š. (2024). Borders and Population Growth: Evidence from a Century of Border Regime Changes on the Austrian-Czech Border. MUNI ECON Working Paper n. 2024-03. Brno: Masaryk University. https://doi.org/10.5817/WP_MUNI_ECON_2024-03 2024-02 Prochazka, J., Pandey, S., Castek, O., Firouzjaeiangalougah, M. (2024). Replication of Changing Hearts and Minds? Why Media Messages Designed to Foster Empathy Often Fail (Gubler et al., 2022). MUNI ECON Working Paper n. 2024-02. Brno: Masaryk University. https://doi.org/10.5817/WP_MUNI_ECON_2024-02 2024-01 Marini, M. M., Ulivieri, G. 2024. Meta-analyses in Economic Psychology: A sustainable approach to cross-cultural differences. MUNI ECON Working Paper n. 2024-01. Brno: Masaryk University. https://doi.org/10.5817/WP_MUNI_ECON_2024-01 2023-09 Levi, E., Ramalingam, A. 2023. Absolute vs. relative poverty and wealth: Cooperation in the presence of between-group inequality. MUNI ECON Working Paper n. 2023-09. Brno: Masaryk University. https://doi.org/10.5817/WP_MUNI_ECON_2023-09 2023-08 Fumarco, L., Harrell, B., Button, P., Schwegman, D., Dils, E. 2023. Gender Identity, Race, and Ethnicity-based Discrimination in Access to Mental Health Care: Evidence from an Audit Correspondence Field Experiment. MUNI ECON Working Paper n. 2023-08. Brno: Masaryk University. https://doi.org/10.5817/WP_MUNI_ECON_2023-08 2023-07 Levi, E., Bayerlein, M., Grimalda, G., Reggiani, T. 2023. Narratives on migration and political polarization: How the emphasis in narratives can drive us apart. MUNI ECON Working Paper n. 2023-07. Brno: Masaryk University. https://doi.org/10.5817/WP_MUNI_ECON_2023-07 2023-06 Fumarco, L., Gaddis, S. M., Sarracino, F., Snoddy, I. 2023. sendemails: An automated email package with multiple applications. MUNI ECON Working Paper n. 2023-06. Brno: Masaryk University. https://doi.org/10.5817/WP_MUNI_ECON_2023-06 2023-05 Harrell, B., Fumarco, L., Button, P., Schwegman, D., Denwood, K. 2023. The Impact of COVID-19 on Access to Mental Healthcare Services. MUNI ECON Working Paper n. 2023-05. Brno: Masaryk University. https://doi.org/10.5817/WP_MUNI_ECON_2023-05 2023-04 Friedhoff, T., Au, C., Krahnhof, P. 2023. Analysis of the Impact of Orthogonalized Brent Oil Price Shocks on the Returns of Dependent Industries in Times of the Russian War. MUNI ECON Working Paper n. 2023-04. Brno: Masaryk University. https://doi.org/10.5817/WP_MUNI_ECON_2023-04 2023-03 Mikula, Š., Reggiani, T., Sabatini, F. 2023. The long-term impact of religion on social capital: lessons from post-war Czechoslovakia. MUNI ECON Working Paper n. 2023-03. Brno: Masaryk University. https://doi.org/10.5817/WP_MUNI_ECON_2023-03 2023-02 Clò, S., Reggiani, T., Ruberto, S. 2023. onsumption feedback and water saving: An experiment in the metropolitan area of Milan. MUNI ECON Working Paper n. 2023-02. Brno: Masaryk University. https://doi.org/10.5817/WP_MUNI_ECON_2023-02 2023-01 Adamus, M., Grežo, M. 2023. Attitudes towards migrants and preferences for asylum and refugee policies before and during Russian invasion of Ukraine: The case of Slovakia. MUNI ECON Working Paper n. 2023-01. Brno: Masaryk University. https://doi.org/10.5817/WP_MUNI_ECON_2023-01 2022-12 Guzi, M., Kahanec, M., Mýtna Kureková, L. 2022. The Impact of Immigration and Integration Policies On Immigrant-Native Labor Market Hierarchies. MUNI ECON Working Paper n. 2022-12. Brno: Masaryk University. https://doi.org/10.5817/WP_MUNI_ECON_2022-12 2022-11 Antinyan, A., Corazzini, L., Fišar, M., Reggiani, T. 2022. Mind the framing when studying social preferences in the domain of losses. MUNI ECON Working Paper n. 2022-11. Brno: Masaryk University. https://doi.org/10.5817/WP_MUNI_ECON_2022-11 2022-10 Corazzini, L., Marini, M. 2022. Focal points in multiple threshold public goods games: A single-project meta-analysis. MUNI ECON Working Paper n. 2022-10. Brno: Masaryk University. https://doi.org/10.5817/WP_MUNI_ECON_2022-10 2022-09 Fazio, A., Scervini, F., Reggiani, T. 2022. Social media charity campaigns and pro-social behavior. Evidence from the Ice Bucket Challenge.. MUNI ECON Working Paper n. 2022-09. Brno: Masaryk University. https://doi.org/10.5817/WP_MUNI_ECON_2022-09 2022-08 Coufalová, L., Mikula, Š. 2022. The Grass Is Not Greener on the Other Side: The Role of Attention in Voting Behaviour.. MUNI ECON Working Paper n. 2022-08. Brno: Masaryk University. https://doi.org/10.5817/WP_MUNI_ECON_2022-08 2022-07 Fazio, A., Reggiani, T. 2022. Minimum wage and tolerance for inequality.. MUNI ECON Working Paper n. 2022-07. Brno: Masaryk University. https://doi.org/10.5817/WP_MUNI_ECON_2022-07 2022-06 Mikula, Š., Reggiani, T. 2022. Residential-based discrimination in the labor market. MUNI ECON Working Paper n. 2022-06. Brno: Masaryk University. https://doi.org/10.5817/WP_MUNI_ECON_2022-06 2022-05 Mikula, Š., Molnár, P. 2022. Expected Transport Accessibility Improvement and House Prices: Evidence from the Construction of the World’s Longest Undersea Road Tunnel. MUNI ECON Working Paper n. 2022-05. Brno: Masaryk University. https://doi.org/10.5817/WP_MUNI_ECON_2022-05 2022-04 Coufalová, L., Mikula, Š., Ševčík, M. 2022. Homophily in Voting Behavior: Evidence from Preferential Voting. MUNI ECON Working Paper n. 2022-04. Brno: Masaryk University. https://doi.org/10.5817/WP_MUNI_ECON_2022-04 2022-03 Kecskésová, M., Mikula, Š. 2022. Malaria and Economic Development in the Short-term: Plasmodium falciparum vs Plasmodium vivax. MUNI ECON Working Paper n. 2022-03. Brno: Masaryk University. https://doi.org/10.5817/WP_MUNI_ECON_2022-03 2022-02 Mladenović, D., Rrustemi, V., Martin, S., Kalia, P., Chawdhary, R. 2022. Effects of Sociodemographic Variables on Electronic Word of Mouth: Evidence from Emerging Economies. MUNI ECON Working Paper n. 2022-02. Brno: Masaryk University. https://doi.org/10.5817/WP_MUNI_ECON_2022-02 2022-01 Mikula, Š., Montag, J. 2022. Roma and Bureaucrats: A Field Experiment in the Czech Republic. MUNI ECON Working Paper n. 2022-01. Brno: Masaryk University. https://doi.org/10.5817/WP_MUNI_ECON_2022-01 2021-14 Abraham, E. D., Corazzini, L., Fišar, M., Reggiani, T. 2021. Delegation and Overhead Aversion with Multiple Threshold Public Goods. MUNI ECON Working Paper n. 2021-14. Brno: Masaryk University. https://doi.org/10.5817/WP_MUNI_ECON_2021-14 2021-13 Corazzini, L., Cotton, C., Longo, E., Reggiani, T. 2021. The Gates Effect in Public Goods Experiments: How Donations Flow to the Recipients Favored by the Wealthy. MUNI ECON Working Paper n. 2021-13. Brno: Masaryk University. https://doi.org/10.5817/WP_MUNI_ECON_2021-13 2021-12 Staněk, R., Krčál, O., Mikula, Š. 2021. Social Capital and Mobility: An Experimental Study. MUNI ECON Working Paper n. 2021-12. Brno: Masaryk University. https://doi.org/10.5817/WP_MUNI_ECON_2021-12 2021-11 Staněk, R., Krčál, O., Čellárová, K. 2021. Pull yourself up by your bootstraps: Identifying procedural preferences against helping others in the presence. MUNI ECON Working Paper n. 2021-11. Brno: Masaryk University. https://doi.org/10.5817/WP_MUNI_ECON_2021-11 2021-10 Levi, E., Sin, I., Stillman, S. 2021. Understanding the Origins of Populist Political Parties and the Role of External Shocks. MUNI ECON Working Paper n. 2021-10. Brno: Masaryk University. https://doi.org/10.5817/WP_MUNI_ECON_2021-10 2021-09 Adamus, M., Grežo, M. 202. Individual Differences in Behavioural Responses to the Financial Threat Posed by the COVID-19 Pandemic. MUNI ECON Working Paper n. 2021-09. Brno: Masaryk University. https://doi.org/10.5817/WP_MUNI_ECON_2021-09 2021-08 Hargreaves Heap, S. P., Karadimitropoulou, A., Levi, E. 2021. Narrative based information: is it the facts or their packaging that matters?. MUNI ECON Working Paper n. 2021-08. Brno: Masaryk University. https://doi.org/10.5817/WP_MUNI_ECON_2021-08 2021-07 Hargreaves Heap, S. P., Levi, E., Ramalingam, A. 2021. Group identification and giving: in-group love, out-group hate and their crowding out. MUNI ECON Working Paper n. 2021-07. Brno: Masaryk University. https://doi.org/10.5817/WP_MUNI_ECON_2021-07 2021-06 Medda, T., Pelligra, V., Reggiani, T. 2021. Lab-Sophistication: Does Repeated Participation in Laboratory Experiments Affect Pro-Social Behaviour?. MUNI ECON Working Paper n. 2021-06. Brno: Masaryk University. https://doi.org/10.5817/WP_MUNI_ECON_2021-06 2021-05 Guzi, M., Kahanec, M., Ulceluse M., M. 2021. Europe’s migration experience and its effects on economic inequality. MUNI ECON Working Paper n. 2021-05. Brno: Masaryk University. https://doi.org/10.5817/WP_MUNI_ECON_2021-05 2021-04 Fazio, A., Reggiani, T., Sabatini, F. 2021. The political cost of lockdown´s enforcement. MUNI ECON Working Paper n. 2021-04. Brno: Masaryk University. https://doi.org/10.5817/WP_MUNI_ECON_2021-04 2021-03 Peciar, V. Empirical investigation into market power, markups and employment. MUNI ECON Working Paper n. 2021-03. Brno: Masaryk University. https://doi.org/10.5817/WP_MUNI_ECON_2021-03 2021-02 Abraham, D., Greiner, B., Stephanides, M. 2021. On the Internet you can be anyone: An experiment on strategic avatar choice in online marketplaces. MUNI ECON Working Paper n. 2021-02. Brno: Masaryk University. https://doi.org/10.5817/WP_MUNI_ECON_2021-02 2021-01 Krčál, O., Peer, S., Staněk, R. 2021. Can time-inconsistent preferences explain hypothetical biases?. MUNI ECON Working Paper n. 2021-01. Brno: Masaryk University. https://doi.org/10.5817/WP_MUNI_ECON_2021-01 2020-04 Pelligra, V., Reggiani, T., Zizzo, D.J. 2020. Responding to (Un)Reasonable Requests by an Authority. MUNI ECON Working Paper n. 2020-04. Brno: Masaryk University. https://doi.org/10.5817/WP_MUNI_ECON_2020-04 2020-03 de Pedraza, P., Guzi, M., Tijdens, K. 2020. Life Dissatisfaction and Anxiety in COVID-19 pandemic. MUNI ECON Working Paper n. 2020-03. Brno: Masaryk University. https://doi.org/10.5817/WP_MUNI_ECON_2020-03 2020-02 de Pedraza, P., Guzi, M., Tijdens, K. 2020. Life Satisfaction of Employees, Labour Market Tightness and Matching Efficiency. MUNI ECON Working Paper n. 2020-02. Brno: Masaryk University. https://doi.org/10.5817/WP_MUNI_ECON_2020-02 2020-01 Fišar, M., Reggiani, T., Sabatini, F., Špalek, J. 2020. a. MUNI ECON Working Paper n. 2020-01. Brno: Masaryk University. https://doi.org/10.5817/WP_MUNI_ECON_2020-01