This collection of essays attempts to address the question: what are the main driving forces of the recent changes in entrepreneurship and business dynamics in the US? In the first chapter, I examine changes within businesses in the US by showing that the share of value added by pass-through entities (sole proprietorships, partnerships, LLCs, and S-corporations) almost doubled while that of C-corporations has declined by one-forth. During this period, there have been notable changes in the tax structure and tax avoidance within these entities. I develop a dynamic growth model with endogenous tax avoidance, occupation choice, and uninsurable entrepreneurial risk to study the extent to which changes in taxation can account for the observed reallocation of output. My model results show that changes in tax structure account for 14 percent of the reallocation of output share observed in the US. I also find that the cumulative effect of changes in taxation, borrowing ability, and tax avoidance accounts for about 26 percent of the reallocation of output. In Chapter 2, I examine a different perspective on business dynamics by documenting the decline in business formation and entrepreneurship in the US. In fact, I document that entrepreneurship is more prevalent in married households and amongmen, and that these groups have undergone a greater decline in entrepreneurship since the 1980s. Additionally, I document that changes in the number of married households and the increase in female labor force participation account for over 40 percent of the overall fall in entrepreneurship in this period. To understand the relationship between demographic composition factors and entrepreneurship, I develop a model with an occupation choice for individuals of different marital status, college skills, and gender. The model takes into account important features of the data, including the extent of marital sorting, the skill premium, the gender wage gap, and the gender business income gap. My results indicate that changes in the demographic composition (share of married households, fraction of skilled individuals, marital sorting) account for 76% of the decline in entrepreneurship, 68.4% of the fall in married entrepreneurs, and 70.5% of the decrease in male entrepreneurs. Moreover, considering all changes account for 82.8% of the observed fall in entrepreneurship.

A central concern for modern macroeconomics is incorporating and understanding worker heterogeneity. The following two essays explore labor market dynamics along the dimensions of worker heterogeneity, search frictions, and policy. In each essay, I construct a macroeconomic model of the labor market, calibrate the model using micro data, and use the model to interpret labor market outcomes and evaluate policy. In the first chapter, I build an equilibrium lifecycle model of wages in which heterogeneous workers endogenously invest in human capital accumulation and on-the-job search effort while firms post jobs. I discipline the model using microdata from the Survey of Income and Program Participation. The calibrated model shows that on-the-job search drives lifecycle wage growth while heterogeneous human capital accumulation drives lifecycle wage dispersion. Then, I use the model as a laboratory to study the effects of tax and transfer progressivity. An increase in progressivity decreases wages, primarily due to reduced on-the-job search effort. Interactions between human capital, search, and job posting amplify the decrease in wages. Surprisingly, an increase in progressivity has little effect on wage dispersion because the effects from the human capital and search channels offset each other. The second chapter deals with the persistence of the unemployment rate over the business cycle. Standard search models contain little internal propagation and predict that, after shocks, the unemployment rate quickly converges to its steady state level. I show that duration dependence in unemployment (the fact that unemployed workers with longer unemployment spells are less likely to find jobs) helps explain the persistence of the unemployment rate. I embed duration dependence in an otherwise standard search model and show that it significantly increases the unemployment rate persistence, reconciling the model to the data. Intuitively, after recessions, the composition of the unemployment pool shifts to the long-term unemployed. Because of duration dependence, the long-term unemployed have lower job finding rates, and the shift in composition decreases the aggregate job finding rate, slowing recovery. The magnitude of the effect depends on the extent to which duration dependence is causal rather than a consequence of worker heterogeneity.

This dissertation studies how forecasting performance can be improved in big data. The first chapter with Seung C. Ahn considers Partial Least Squares (PLS) estimation of a time-series forecasting model with data containing a large number of time series observations of many predictors. In the model, a subset or a whole set of the latent common factors in predictors determine a target variable. First, the optimal number of the PLS factors for forecasting could be smaller than the number of the common factors relevant for the target variable. Second, as more than the optimal number of PLS factors is used, the out-of-sample explanatory power of the factors could decrease while their in-sample power may increase. Monte Carlo simulation results also confirm these asymptotic results. In addition, simulation results indicate that the out-of-sample forecasting power of the PLS factors is often higher when a smaller than the asymptotically optimal number of factors are used. Finally, the out-of-sample forecasting power of the PLS factors often decreases as the second, third, and more factors are added, even if the asymptotically optimal number of the factors is greater than one. The second chapter studies the predictive performance of various factor estimations comprehensively. Big data that consist of major U.S. macroeconomic and finance variables, are constructed. 148 target variables are forecasted, using 7 factor estimation methods with 11 information criteria. First, the number of factors used in forecasting is important and Incorporating more factors does not always provide better forecasting performance. Second, using consistently estimated number of factors does not necessarily improve predictive performance. The first PLS factor, which is not theoretically consistent, very often shows strong forecasting performance. Third, there is a large difference in the forecasting performance across different information criteria, even when the same factor estimation method is used. Therefore, the choice of factor estimation method, as well as the information criterion, is crucial in forecasting practice. Finally, the first PLS factor yields forecasting performance very close to the best result from the total combinations of the 7 factor estimation methods and 11 information criteria.

This dissertation consists of three essays on the task approach to labor markets. In the first chapter, I document that since 2000 the polarization of wages in the U.S. labor market stopped, as the wages of non-routine manual occupations fell in relative and absolute terms. I analyze the end of wage polarization through the lens of a dynamic general equilibrium model with occupation-biased technical change, human capital accumulation, and occupational mobility. I show that wage polarization ended because workers in non-routine manual occupations had lower initial human capital and lower human capital accumulation over time, and because after 2000 mobility across occupations fell, which magnified the differences in human capital accumulation across occupations. The second chapter estimates the effect of the import competition from China on the intensity of tasks performed by workers within U.S. manufacturing establishments between 2002 and 2017. I measure the changes in the intensity of these tasks by linking information on occupational employment from the Occupational Employment Statistics to the occupational characteristics from the Occupational Information Network (O*NET). I find that this “China shock” led establishments to significantly decrease the intensity of cognitive and interpersonal tasks, and to increase the intensity of manual and routine tasks. These estimations are consistent with US establishments reallocating employment to become more similar to their Chinese competitors and have important implications for the design of public policies. The third chapter explores the importance of changes in the intensity of tasks performed by workers to explain the evolution of wages. Despite changes in the workplace, the literature is based on the questionable assumption that the intensity of tasks remains constant over time. I harmonize and compare over time the intensity of non-routine cognitive, non-routine manual, interpersonal, and routine tasks in the Dictionary of Occupation Title (DOT) and the O*NET. I find the new fact that a sizable part of wage changes is due to increases in the return and the intensity of cognitive tasks. I show that this fact has implications for three well-documented wage trends during the last decades: wage polarization, increasing college premium, decreasing gender-wage gap.

These essays are my attempt to answer a big picture question in economics "why some countries are richer than others?". In the first chapter, I document that for a group of 38 countries ranging from low to high income, managers in richer countries are more skilled, and the relative income of managers to non-managers along with skill premium is lower in richer countries. I use a model of investment in skills and occupational choice in which countries differ in productivity level and size-dependent distortions. I find that exogenous productivity differences alone can produce the abovefacts qualitatively, but size-dependent distortions are needed to account for these facts quantitatively. Chapter two accounts for the sources of world productivity growth, using data for more than 36 industries and 40 economies. Productivity growth in advanced economies slowed but emerging markets grew more quickly, which kept global productivity growth relatively constant until 2010. World productivity growth is highly volatile from year to year, which primarily reflects shifts in the reallocation of labor. Deviations from Purchasing Power Parity account for about a third of the shifts. Though markups are large and rise over time, they only modestly affect measured industry-level productivity growth. In chapter three, I document that the mean and dispersion of pre-tax labor earnings grow faster over the life-cycle in the U.S. than in some European countries and individuals with at least a college degree are key for these facts. I use a life-cycle model of human capital accumulation and elastic labor supply which features non-linear taxation and a college choice and investments during college. The model economy is consistent with earnings distribution among college and non-college individuals in the U.S. Non-linear taxation suppresses pre-tax earnings, reduces college attendance and investments during college. More generous subsidies for college exacerbate labor earning inequality. Differences in taxation and college subsidies account for 94% of the differences in mean earnings, and 80% of the differences in inequality over the life-cycle across the U.S. and European countries.

This dissertation consist two chapters related with misallocation and economic development.

The first chapter studies the organization of production, as summarized by the number of managers per plant, the number of workers per manager and the mean size of plants in terms of employment. First, I document that in the manufacturing sector, richer countries tend to have (i) more managers per plant, (ii) less workers per manager and (iii) larger plants on average. I then extend a knowledge-based hierarchies model of the organization of production where the communication technology depends on the managerial level in the hierarchy and the abilities of subordinates. I estimate model parameters so that the model jointly produces plant size distribution and number of managers per plant in the United States manufacturing sector. I find that when the largest, more complex, plants face distortions that are twice as large as distortions faced by smaller plants, output declines by 33.4% and the number of managers per plant falls by 30%. Moreover, I find that a 10% increase in communication cost parameters can account for a 35% decrease in the aggregate output without having a significant effect on the number of managers per plant.

The second chapter examines the relationship between bribery, plant size and economic development. Using the Enterprise Survey, I document that small plants spend higher fraction of their output on bribery than big plants do. Then I develop a one sector growth model in which size-dependent distortions, bribery opportunities and different plant sizes coexist. I find that size-dependent distortions become less distortionary in the presence of bribery opportunities and the effect of such distortions on the plant size become reversed since bigger plants are able to avoid from distortions by paying larger bribes. My results indicate that changes in the distortion level do not affect output and size significantly because managers are able to circumvent the distortions by adjusting their bribery expenditures. However, the removal of distortions can have a substantial effect on both the output and the mean size. Output in Turkey can increase by 12.3%, while the mean size can increase by almost double.

This dissertation consists of two essays with a macroeconomic approach to economic development. These essays explore specific barriers that prevent economic agents from exploiting opportunities across regions or sectors in developing countries, and to what extent the observed allocations are inefficient outcomes or just an efficient response to economic fundamentals and technological constraints.

The first chapter is motivated by the fact that a prominent feature of cities in developing countries is the existence of slums: locations with low housing-quality and informal property rights. This paper focuses on the allocation of land across slums and formal housing, and emphasizes the role of living in central urban areas for the formation of slums. I build a quantitative spatial general equilibrium model to study the aggregate effects of anti-slum policies and use microdata from India for the quantitative implementation. According to my findings, demolishing slums in central urban areas leads to a decrease in welfare, aggregate labor productivity, and urban population. In contrast, decreasing formal housing distortions in India to the U.S. level increases the urban population share by 20% and labor productivity by 2.4%, and reduces the share of the urban population living in slums by 19%.

The second chapter is motivated by the fact that labor productivity gaps between rich and poor countries are much larger for agriculture than for non-agriculture. Using detailed data from Mexican farms, this paper shows that value added per worker is frequently over two times larger in cash crops than in staple crops, yet most farmers choose to produce staples. These findings imply that the agricultural productivity gap is actually a staple productivity gap and understanding production decisions of farmers is crucial to explain why labor productivity is so low in poor countries. This paper develops a general equilibrium framework in which subsistence consumption and interregional trade costs determine the efficient selection of farmers into types of crops. The quantitative results of the model imply that decreasing trade costs in Mexico to the U.S. level reduces the ratio of employment in staple to cash crops by 17% and increases agricultural labor productivity by 14%.