Dissertations

Taxation impacts social welfare in an intricate manner.  Currently employed tax instruments throughout the world possess inherently different salience characteristics.  Tax salience effect refers to the optimization error occurring when agents do not fully account for taxes that do not appear in posted prices.  Purchase size effect (PSE), refers to how the tax salience effect changes based on the size of the monetary stake of the purchase.  The first chapter of this dissertation aims to assess whether participants in a ‘quasi-field’ laboratory experiment with an online shopping environment, real goods, low and high price/type goods, and with conspicuous taxation cues will make tax salience “errors” and if they will make less of them when purchasing more expensive goods.  This is the first experiment in the literature to focus on the PSE dimension of tax salience and to divide participants into low and high income-/wealth-analogous budget levels. 

The results confirm consistent and significant tax salience effects and reveal the PSE despite the cues.  Tax salience effect is most driven by high-budget participants while low-budget participants exhibit several more significant PSE difference estimates.  Such a combination suggests that those with more constraining budgets were less likely to be making a tax salience error generally and even lesser likely with more expensive goods, even though the high-budget group faced higher utility stakes.  This may imply the dominance of one of the budget adjustment mechanisms, which is predicted to lead to a positive social welfare outcome. 

Supplemental, informative findings indicate that these behaviors are not gender specific and appear to be an intentional/strategic choice, substantially mediated by one’s budget constraints and shopping speed.  These results suggest that rational inattention drives the tax salience effect.  With correlation to income/wealth measures, these findings imply an innate progressivity in less salient consumption tax instruments.

The second chapter examines the presence of parental preference for a particular gender of children.  We test for the predominance of the two main explanations for the existence of such preference, namely differences in the costs of raising sons and daughters versus the gender bias (corresponding to parental utility derived from a child’s gender or from characteristics exclusive to that gender).  First, we use recent EU-SILC data from several Balkan and Scandinavian countries to confirm that the gender of the firstborn predicts the likelihood of a given family having three children or more—a common measure of parental gender preference.  We confirm son preference in certain Balkan countries and daughter preference in Scandinavian countries.  Both having a first child of the preferred gender and of the more costly gender can decrease the probability of having three or more children because parents may already be content or may lack sufficient resources, respectively.  Next, we use information on household consumption to differentiate the two explanations.  We argue that under the differential cost hypothesis, parents of children of the more costly gender should spend more on goods for children and less on household public goods as well as on parental personal consumption.  In contrast, having children of the preferred gender should increase spending on household public goods since such families have higher marriage surplus and are more stable.  Our evidence corroborates the cost difference explanation in countries exhibiting daughter preference.

The third chapter evaluates a one-time, immediate policy initiative enacted by the Republic of Georgia that shifted public office working hours from 10:00-19:00 to 9:00-18:00, which affected the work schedules of government employees.  Although the policy affected approximately 200,000 employees, it has never been evaluated; and to our knowledge, nor has any similar policy in any economic literature.  The effects of the policy impact gender and family types asymmetrically, relating this paper most closely to work-family conflict literature, which provides a framework for making two opposing predictions based on gender similarity and difference models.  Furthermore, we examine how the policy impacts gender inequality through female labor participation.

Given that the policy did not affect the private sector, we employ a difference-in-differences approach using the National Statistics Office of Georgia Households Incomes and Expenditures Survey from 2013-2016.  We find that the policy primarily produces a significant reduction in the average level of working hours of full-time employees with children, directly in line with the prediction following the gender similarity model.  We also find a significant increase in average work hour engagement by women without children.  However, the placebo effect analysis identifies this as an already existing trend and the short-term analysis indicates that this is an ordinal reaction to the reduction of engagement by full-time employees with children.  We conclude that this increase is a secondary, indirect effect and that the policy did not directly cause an increase in female labor participation.  Furthermore, since men with children were most negatively affected and women picked up the gains, the policy may also indirectly increase overall gender equality.

Full Text: "Explorations into Behavioral Phenomena"