Discussion Paper No. 424
September 11, 2023
Abadie's Kappa and Weighting Estimators of the Local Average Treatment Effect
Author:
Abstract:
In this paper we study the finite sample and asymptotic properties of various weighting estimators of the local average treatment effect (LATE), several of which are based on Abadie's (2003) kappa theorem. Our framework presumes a binary treatment and a binary instrument, which may only be valid after conditioning on additional covariates. We argue that one of the Abadie estimators, which is weight normalized, is preferable in many contexts. Several other estimators, which are unnormalized, do not generally satisfy the properties of scale invariance with respect to the natural logarithm and translation invariance, thereby exhibiting sensitivity to the units of measurement when estimating the LATE in logs and the centering of the outcome variable more generally. On the other hand, when noncompliance is one-sided, certain unnormalized estimators have the advantage of being based on a denominator that is bounded away from zero. To reconcile these findings, we demonstrate that when the instrument propensity score is estimated using an appropriate covariate balancing approach, the resulting normalized estimator also shares this advantage. We use a simulation study and three empirical applications to illustrate our findings. In two cases, the unnormalized estimates are clearly unreasonable, with ``incorrect'' signs, magnitudes, or both.
Keywords:
JEL-Classification:
Download:
Discussion Paper No. 423
Peer Effects Heterogeneity and Social Networks in Education
Author:
Abstract:
This study focuses on the role of heterogeneity in network peer effects by accounting for network-specific factors and different driving mechanisms of peer behavior. We propose a novel Multivariate Instrumental Variable (MVIV) estimator which is consistent for a large number of networks keeping the individual network size bounded. We apply this approach to estimate peer effects on school achievement exploiting the network structure of friendships within classrooms. The empirical evidence presented is based on a unique network dataset from German upper secondary schools. We show that accounting for heterogeneity is not only crucial from a statistical perspective, but also yields new structural insights into how class size and gender composition affect school achievement through peer behavior.
Keywords:
network heterogeneity; peer effects; multivariate instrumental variables; minimum distance estimation; school achievement;
JEL-Classification:
D85; L14; I21; C30; C36;
Download:
Discussion Paper No. 422
Bit by Bit: Colocation and the Death of Distance in Software Developer Networks
Author:
Abstract:
Digital work settings potentially facilitate remote collaboration and thereby decrease geographic frictions in knowledge work. Here, I analyze spatial collaboration patterns of some 191 thousand software developers in the United States on the largest code repository platform GitHub. Despite advanced digitization in this occupation, developers are geographically highly concentrated, with 79.8% of users clustering in only ten economic areas, and colocated developers collaborate about nine times as much as non-colocated developers. However, the colocation effect is much smaller than in less digital social or inventor networks, and apart from colocation geographic distance is of little relevance to collaboration. This suggests distance is indeed less important for collaboration in a digital work setting while other strong drivers of geographic concentration remain. Heterogeneity analyses provide insights on which types of collaboration tend to colocate: the colocation effect is smaller within larger organizations, for high-quality projects, among experienced developers, and for sporadic interactions. Overall, this results in a smaller colocation effect in larger economic areas.
Keywords:
geography; digitalization; networks; knowledge economy; colocation;
JEL-Classification:
L84; O18; O30; R32;
Download:
Discussion Paper No. 421
Automation and Polarization
Author:
Abstract:
We develop an assignment model of automation. Each of a continuum of tasks of variable complexity is assigned to either capital or one of a continuum of labor skills. We characterize conditions for interior automation, whereby tasks of intermediate complexity are assigned to capital. Interior automation arises when the most skilled workers have a comparative advantage in the most complex tasks relative to capital, and because the wages of the least skilled workers are sufficiently low relative to their productivity and the effective cost of capital in low-complexity tasks. Minimum wages and other sources of higher wages at the bottom make interior automation less likely. Starting with interior automation, a reduction in the cost of capital (or an increase in capital productivity) causes employment and wage polarization. Specifically, further automation pushes workers into tasks at the lower and upper ends of the task distribution. It also monotonically increases the skill premium above a skill threshold and reduces the skill premium below this threshold. Moreover, automation tends to reduce the real wage of workers with comparative advantage profiles close to that of capital. We show that large enough increases in capital productivity ultimately induce a transition to low-skill automation and qualitatively alter the effects of automation - thereafter inducing monotone increases in skill premia rather than wage polarization.
Keywords:
assignment; automation; inequality; polarization; tasks; wages;
JEL-Classification:
J23; J31; O33;
Download:
Discussion Paper No. 420
August 31, 2023
Redistributive Income Taxation with Directed Technical Change
Author:
Abstract:
What are the implications of (endogenous) directed technical change for the design of redistributive income taxes? I study this question in a Mirrleesian economy augmented to include endogenous technology development and adoption choices by firms. Under certain conditions, any progressive tax reform induces technical change that compresses the pre-tax wage distribution. The key intuition is that progressive tax reforms tend to increase labor supply of less skilled relative to more skilled workers, which induces firms to develop and use technologies that are more complementary to the less skilled. These directed technical change effects make the optimal tax scheme more progressive, raising marginal tax rates at the right tail of the income distribution and lowering them at the left tail. For reasonable calibrations, the impact of directed technical change on the optimal tax is quantitatively important: optimal marginal tax rates are reduced substantially for incomes below the median and increase monotonically over the bulk of the income distribution instead of being U-shaped (as in most of the previous literature).
Keywords:
optimal taxation; directed technical change; endogenous technical change; wage inequality;
JEL-Classification:
H21; H23; H24; J31; O33;
Download:
Discussion Paper No. 419
August 25, 2023
Allegations of Sexual Misconduct, Accused Scientists, and Their Research
Author:
Abstract:
Does the scientific community sanction sexual misconduct? Using a sample of scientists accused of sexual misconduct at US universities, we find that their prior work is cited less after allegations surface. The effect weakens with distance in the coauthorship network, indicating that researchers learn about allegations through their peers. Among the closest peers, male authors react more strongly, suggesting that they feel a greater need to disassociate themselves from the accused. In male-dominated fields, the effects on citations are more muted. Accused scientists are more likely to leave academic research, to move to non-university institutions, and to publish less.
Keywords:
sexual misconduct; scientific community; scientific impact;
JEL-Classification:
J16; M14; I23; K4;
Download:
Discussion Paper No. 418
Structural Shocks and Political Participation in the US
Author:
Abstract:
This paper examines the impact of the large structural shocks -- automation and import competition -- on voter turnout during US federal elections from 2000 to 2016. Although the negative income effect of both shocks is comparable, we find that political participation decreases significantly in counties more exposed to industrial robots. In contrast, the exposure to rising import competition does not reduce voter turnout. A survey experiment reveals that divergent beliefs about the effectiveness of government intervention drive this contrast. Our study highlights the role of beliefs in the political economy of technological change.
Keywords:
automation; trade; labor demand; voter turnout;
JEL-Classification:
D72; J23; F16;
Download:
Discussion Paper No. 417
August 23, 2023
The Behavioral Additionality of Government Research Grants
Author:
Abstract:
There are different forms of public support for industrial R&D. Some attempt to increase innovation by prompting firms to undertake more challenging projects than they would otherwise do. Access to a dataset from one such program, the Austrian Research Promotion Agency, allows me to examine the effect of research grants on firms' patenting outcomes. My estimates suggest that a government research grant increases the propensity to file a patent application with the European Patent Office by around 12 percentage points. Stronger effects appear for more experienced firms of advanced age. Additional evidence indicates that grants induce experienced firms to develop unconventional patents and patents that draw on knowledge novel to the firm. I interpret the findings in a "exploration vs. exploitation" model, in which grants are targeted at ambitious projects that face internal competition from more conventional projects within firms. The model shows that this mechanism is more salient in experienced firms, leading to a stronger response in behavior for this group of firms.
Keywords:
JEL-Classification:
O38;
Download:
Discussion Paper No. 416
August 10, 2023
Complementing Business Training with Access to Finance: Evidence from SMEs in Kenya
Author:
Abstract:
In this paper, we study the complementarity between business training and access to financial capital for small and medium enterprises (SMEs) in Kenya. All participants in a business training program are offered training. One-third of participants are offered loans immediately after training (Concurrent Loan group), one-third are offered loans six weeks after training (Delayed Loan group), and the remaining third are offered loans after another four weeks (Control group). While a long delay between training and loans may reduce knowledge retention and application by SMEs in the presence of complementarity, concurrent access to loans and associated business spending may crowd out the entrepreneurs' attention from improving business practices. We find evidence for the latter in both intention-to-treat and treatment-on-the-treated estimates. While SMEs in both Control and Delayed Loan groups improve their business practices, SMEs in the Concurrent Loan group who take loans do not improve their practices at all. Moreover, entrepreneurs who take loans spend less time on their businesses and their business revenue falls. Our evidence is consistent with the entrepreneurs in our study using loans to substitute for their income.
Keywords:
business training; access to finance;
JEL-Classification:
O12; L26; M53;
Download:
Discussion Paper No. 415
Whom to Inform about Prices? Evidence from the German Fuel Market
Author:
Abstract:
Combining a theoretical model of imperfect information with empirical evidence, we show how the effect of providing price information to consumers depends on how well informed they are beforehand. Theoretically, an increase in consumer information decreases prices more, the fewer ex ante informed consumers there are. Empirically, we study mandatory price disclosure in the German fuel market for two fuel types that differ in ex ante consumer information. The decline in prices is stronger when there are fewer ex ante informed consumers. The magnitude of the treatment effect declines over time but is intensified by local follow-on information campaigns.
Keywords:
mandatory price disclosure; consumer information; retail fuel market;
JEL-Classification:
D83; L41;