Costas (Maks) reset

Question 1

Title: Katz & Murphy (1992)

Answer 1

Changes in Relative Wages, 1963-1987: Supply and Demand Factors

Question 2

Motivation: Katz & Murphy (1992) - Changes in Relative Wages, 1963-1987: Supply and Demand Factors

Answer 2

US wage gaps between high and low skill workers exploded after the late-1970s. Are those gaps mainly due to a flood of college grads (supply) or to rising employer demand for skill (skill-biased technological change)?

Question 3

Setting: Katz & Murphy (1992) - Changes in Relative Wages, 1963-1987: Supply and Demand Factors

Answer 3

United States, 1963-87. Annual March Current Population Survey files track wages and hours for the whole workforce

Question 4

Supply-Only Test: Katz & Murphy (1992) - Changes in Relative Wages, 1963-1987: Supply and Demand Factors

Answer 4

Compute the dot-product of wage changes and quantity changes for every five-year window. If only supply moved, the product should be negative; a positive sign means demand must have shifted, as supply cant explain the shift (hapened in mid 80s)

Question 5

Test Demand: Katz & Murphy (1992) - Changes in Relative Wages, 1963-1987: Supply and Demand Factors

Answer 5

Collapse labour into “high-skill” and “low-skill” productivity units using the constant elasticity substitution (CES) production function.

Estimate the substitution elasticity (how fast or how slow do high skill workers substitute low skill workers) and wage ratio, to get the backed out demand trend (interpreted as skill-biased tech change).

Question 6

Decomposing Demand Shift: Katz & Murphy (1992) - Changes in Relative Wages, 1963-1987: Supply and Demand Factors

Answer 6

Build counterfactual indices to ask how much of the demand shift comes within-industry up-skilling, between-industry reallocation, or trade.

Question 7

Robustness: Katz & Murphy (1992) - Changes in Relative Wages, 1963-1987: Supply and Demand Factors

Answer 7

CES results unchanged when adding union density, import penetration, or alternative industry splits; trade-only and sectoral reallocation stories explain at most a minor share of the overall demand shift.

Technology, and not trade flows, explain the extra skill demand

Question 8

Results: Katz & Murphy (1992) - Changes in Relative Wages, 1963-1987: Supply and Demand Factors

Answer 8

After 1980, there is a tech-driven rise in demand for college-level skill, overrunning earlier supply booms and pushing the wage premium up again. Most of the extra demand appears within industries, consistent with technology-driven up-skilling (SBTC)

Question 9

Title: Autor, Levy & Murnane (2003)

Answer 9

The Skill Content of Recent Technological Change: An Empirical Exploration

Question 10

Motivation: Autor, Levy & Murnane (2003) - The Skill Content of Recent Technological Change: An Empirical Exploration

Answer 10

Since the 1970s, the “middle” of the U.S. job market has thinned out: office clerks and factory workers have become scarcer, while well-paid professional jobs have multiplied. This paper asks whether the spread of cheap computers is the reason.

Question 11

Setting: Autor, Levy & Murnane (2003) - The Skill Content of Recent Technological Change: An Empirical Exploration

Answer 11

US population survey data from 1960-1998 are merged with occupation-level routine scores (from the Dictionary of Occupational Titles)

Question 12

Model: Autor, Levy & Murnane (2003) - The Skill Content of Recent Technological Change: An Empirical Exploration

Answer 12

The production function consists of (routine + computers) and non-routine labour.

Routine labour and computers are perfect substitutes (price of routine labour = price of computers) because employers are profit maximising and rational (i.e. if computers are cheaper, no one will hire routine labour, and vice versa).

When computers get cheaper, workers get displaced and move from routine to non-routine tasks.

Question 13

Empirics: Autor, Levy & Murnane (2003) - The Skill Content of Recent Technological Change: An Empirical Exploration

Answer 13

1) Regress the change in the share of routine tasks on the computer adoption rate to see whether more routine-intensive industries actually switched to computers faster, as per the predictions of the model.

2) Run the same specification separately for college and non-college workers to verify that job routiness is moving for all education levels.

3) Within-occupation tests confirm that tasks inside job titles re-balance as computers get widespread and sectors shed routine tasks.

Question 14

Robustness: Autor, Levy & Murnane (2003) - The Skill Content of Recent Technological Change: An Empirical Exploration

Answer 14

Common computer-price shock applies to all sectors; placebo decades show no effect; identical patterns appear across industries, education groups and occupations

Question 15

Results: Autor, Levy & Murnane (2003) - The Skill Content of Recent Technological Change: An Empirical Exploration

Answer 15

Routine-heavy industries adopt computers earlier, automate clerical/assembly work and expand analytic and interactive tasks — both between and within occupations. This task re-allocation explains the bulk of the post-1970 surge in demand for college-educated labour.

Question 16

Title: Autor & Dorn (2013)

Answer 16

The Growth of Low-Skill Service Jobs and the Polarization of the US Labor Market

Question 17

Motivation: Autor & Dorn (2013) - The Growth of Low-Skill Service Jobs and the Polarization of the US Labor Market

Answer 17

Since 1980 the U.S. job distribution has formed a U-shape: middle-wage clerical posts shrank, while both high-pay professional roles and low-pay service jobs grew. The paper asks whether cheap computers, by automating routine office and production work, pushed displaced workers into in-person services

Question 18

Setting: Autor & Dorn (2013) - The Growth of Low-Skill Service Jobs and the Polarization of the US Labor Market

Answer 18

The authors follow millions of workers in every Census from 1950 to 2005 in the United States

Question 19

Model: Autor & Dorn (2013) - The Growth of Low-Skill Service Jobs and the Polarization of the US Labor Market

Answer 19

2-sector model of production: goods & services. The goods sector combines (routine labour + computers) and abstract non-routine labour. The service sector only includes manual non-routine labour.

If computers get cheaper, firms substitute routine labour for computers, lowering the price of the inputs of production and making goods cheaper as well.

Seeing that goods are cheaper, consumers spend less money on goods and more on services (as goods and services are imperfect substitutes). This incentivises service-firms to hire the displaced routine labour to meet the rising demand.

As such, cheap computers displace routine labour toward manual non-routine labour and raise wages at the top (abstract labour).

Question 20

Empirics: Autor & Dorn (2013) - The Growth of Low-Skill Service Jobs and the Polarization of the US Labor Market

Answer 20

1) Regress changes in employment in the service sector on routine intensity (RTI share) to see if routine-heavy commuting zones (with high RTI) were the fastest to adopt computers and displace routine labour into manual non-routine labour.

2) Within the same commuting zones, see whether the U-shaped polarization is steeper in intitially routine-intensive areas.

3) Regress changes in the share of college students on routine intensity (RTI share) to see if routine-heavy commuting zones attract more college students (abstract non-routine / high-skill workers).

Question 21

Robustness: Autor & Dorn (2013) - The Growth of Low-Skill Service Jobs and the Polarization of the US Labor Market

Answer 21

Instrument the 1980 routine share with the 1950 mix to avoid mechanical bias; control for entity and time fixed effects; replace PC use with hardware-investment data and obtain similar results.

Question 22

Results: Autor & Dorn (2013) - The Growth of Low-Skill Service Jobs and the Polarization of the US Labor Market

Answer 22

Routine-heavy labour markets adopted PCs earlier, then shed clerical and factory jobs while expanding low-skill, face-to-face services.
Within these markets wages stretched into a pronounced U-shape, with it being steepest in high-routine CZs.

The inflow of college-educated workers grew, signalling stronger demand for abstract, non-routine tasks.

Question 23

Title: Goos et al. (2014)

Answer 23

Explaining Job Polarization: Routine-Biased Technological Change and Offshoring

Question 24

Motivation: Goos et al. (2014) - Explaining Job Polarization: Routine-Biased Technological Change and Offshoring

Answer 24

Why have middle-wage jobs vanished while both well-paid professional and low-paid service jobs expanded (polarization)? Authors test whether automation of routine work and its offshoring abroad can jointly reproduce this polar shape. (Many industries, allows for offshoring, Europe rather than US & within or between industires)

Question 25

Setting: Goos et al. (2014) - Explaining Job Polarization: Routine-Biased Technological Change and Offshoring

Answer 25

EU Labour Force Survey, 1993-2010, sixteen West-European countries. Ranked by Routine Task Intensity (RTI—share of repetitive rule-based duties) and by an offshorability index (likelihood tasks can be done remotely).

Question 26

Methodology: Goos et al. (2014) - Explaining Job Polarization: Routine-Biased Technological Change and Offshoring

Answer 26

Main OLS: Regress change in hours worked on RTI, offshorability and the interaction between RTI and offshorability, controlling for fixed effects. Re-estimate with GMM to tackle serial correlation between shrinking occupations and investment in automation. Decomposition isolates within-industry versus between-industry contributions.

Question 27

Robustness: Goos et al. (2014) - Explaining Job Polarization: Routine-Biased Technological Change and Offshoring

Answer 27

Uses clustered-by-occupation SEs, country-specific time trends, alternative RTI scales, person-weight versus hours-weight, exclude surveys that use different standards. Coefficients stable in leave-one-country-out jackknifes.

Question 28

Results: Goos et al. (2014) - Explaining Job Polarization: Routine-Biased Technological Change and Offshoring

Answer 28

High RTI tasks (routine labour) shrink every year/shed employment. 60% of this polarization happens within, and 40% between industries (almost equal effects) - routine biased technological change. It's EU wide Offshoring is not significant once controlling for everything, Thus in routine-heavy industries, firms will prefer to invest in capital to meet the new demand, instead of labour.

Question 29

Title: Deming (2017)

Answer 29

The Growing Importance of Social Skills in the Labor Market

Question 30

Motivation: Deming (2017) - The Growing Importance of Social Skills in the Labor Market

Answer 30

As computers displaced routine cognitive work, returns to pure analytical skills plateaued. Hypothesis: tasks requiring social interaction resist automation, so labour-market demand—and pay—pivot toward social skills, especially when combined with cognitive ability.

Question 31

Setting: Deming (2017) - The Growing Importance of Social Skills in the Labor Market

Answer 31

U.S. Census data 1980-2012, O*NET task ratings for social and cognitive content, wages and measured skills between 79-97 cohorts.

Question 32

Task Trading: Deming (2017) - The Growing Importance of Social Skills in the Labor Market

Answer 32

Workers differ in which micro-tasks they’re quickest at. They gain by swapping tasks (specialisation), yet each swap loses time. These losses shrink when a worker's communication skills are higher (Iceberg costs). As computers automate the routine middle of the labour market, greater variance (abstract) tasks are left, and swapping these uneven tasks is more important and yields bigger gains. Therefore, specialisation requires more task trading, implying more social skills. When paired with cognitive skills, this results in even higher pay (smart workers specialise and collaborate smoothly).

Question 33

Empirical Approach: Deming (2017) - The Growing Importance of Social Skills in the Labor Market

Answer 33

Wages are regressed on cognitive skill, social skill and the interaction between cognitive and social skills. A positive coefficient on the interaction shows that brains pay more when paired with people-skills.

Question 34

Robustness: Deming (2017) - The Growing Importance of Social Skills in the Labor Market

Answer 34

Standard errors clustered by occupation; results unchanged when adding computer-capital, offshoring exposure, union decline, local-labour-market effects. Triple-difference falsification shows no rising premium in cognitive-only roles. IV using sibling fixed effects confirms skill measures not driven by unobserved ability.

Question 35

Results: Deming (2017) - The Growing Importance of Social Skills in the Labor Market

Answer 35

1) Social skills have higher wage returns & pay more, especially when paired with cognitive skills. Social premium doubles from 1980s to 2012 2) High social skill workers re-sort away from the routine jobs and pivot toward interactive, social ones 3) Employment and wages surge in jobs that score high in social and cognitive skills, stagnated in jobs that score high in cognitive skills and low in social skills, and shrank in cognitive only jobs.

Question 36

Title: Cavounidis et al. (2024)

Answer 36

The Nature of Technological Change, 1960-2016

Question 37

Motivation: Cavounidis et al. (2024) - The Nature of Technological Change, 1960-2016

Answer 37

Authors ask which of the 5 basic skills experienced faster or slower productivity growth, and whether those skill-specific shocks can explain between-occupation worker reallocation and the within-occupation re-mix of tasks

Question 38

Setting: Cavounidis et al. (2024) - The Nature of Technological Change, 1960-2016

Answer 38

United States, 1960-2016 occupational data. Five aggregate skills analysed: routine-cognitive, dexterous routine-manual, basic manual, abstract reasoning, social interaction.

Question 39

Methodology: Cavounidis et al. (2024) - The Nature of Technological Change, 1960-2016

Answer 39

Treat every job as a five-element skill mix (summing up to 1). Regress each job’s decade-to-decade employment change on each skill's share of that job. + industry FEs The Estimated Beta Coefficient can be interpreted as relative (relative to the average) skill-specific productivity growth: it reveals which skill technologies raced ahead (negative sign) or lagged behind (positive sign) compared to the average. See whether 5 skills changes are happening within jobs or across occupations. (jobs changing or tasks within jobs?) Also, see cross-derivatives: whether increasing technology skills and skills themselves are substitutes or complements to each other (manual tech -> job uses less manual skill. also cross-tabulations of skill and tech) Use those cross-derivatives plus estimated ΔlnA to predict the within-occupation skill shifts and compare them to the actual DOT/O*NET shifts

Question 40

Robustness: Cavounidis et al. (2024) - The Nature of Technological Change, 1960-2016

Answer 40

Estimates robust to alternative occupation aggregation, dropping recession years, hours vs heads weighting, and augmenting model with occupation-specific capital trends. Out-of-sample validation: skill-growth estimates from 1960-83 accurately forecast 1995-2016 within-occupation task shifts.

Question 41

Results: Cavounidis et al. (2024) - The Nature of Technological Change, 1960-2016

Answer 41

Fastest productivity gains occur in routine-cognitive and finger-dexterous tasks, pushing workers out of those jobs and reducing such task shares those jobs. Slowest productivity gains in abstract and social skills increase their intra-job shares and preserve associated employment. Framework reconciles both the cross-industry polarization of employment and the rising share of interactive and problem-solving tasks.

Question 42

Title: Acemoglu & Restrepo (2022)

Answer 42

Tasks, Automation, and the Rise in U.S. Wage Inequality

Question 43

Motivation: Acemoglu & Restrepo (2022) - Tasks, Automation, and the Rise in U.S. Wage Inequality

Answer 43

Measure displacement of wages (ripple effects / hydrostatics) by automation across industries, linking it to wage and employment hours changes and quantifying how much of rising inequality automation can explain.

Question 44

Data / Setting: Acemoglu & Restrepo (2022) - Tasks, Automation, and the Rise in U.S. Wage Inequality

Answer 44

49 U.S. industries, 1987-2016. Industry-level capital type count.

Question 45

Task Displacement Index: Acemoglu & Restrepo (2022) - Tasks, Automation, and the Rise in U.S. Wage Inequality

Answer 45

Sums, across industries, how much automation cut routine labour times the group’s initial exposure to those tasks.

Question 46

Reduced-form test: Acemoglu & Restrepo (2022) - Tasks, Automation, and the Rise in U.S. Wage Inequality

Answer 46

Reduced form test captures the first hit - how automation lowers wages They regress each group’s real wage change on its task-displacement index while controlling for industry trends and a rich set of group fixed effects. The coefficient shows how much losing one percentage point more of your tasks to automation lowers wages.

Question 47

General-equilibrium model: Acemoglu & Restrepo (2022) - Tasks, Automation, and the Rise in U.S. Wage Inequality

Answer 47

The general Equilibrium model explains the ripple effects of reduced form test. The same task-allocation logic is embedded in a multi-sector model that lets automation in one industry spill into others through labour re-sorting and price changes. Fitting the model to the industry data allows the authors to simulate the full distributional impact of automation (ripple effects & hydrostatics).

Question 48

Robustness: Acemoglu & Restrepo (2022) - Tasks, Automation, and the Rise in U.S. Wage Inequality

Answer 48

Results hold when adding controls for de-unionisation, China import shock, industry mark-ups, and classic skill-bias variables. Split-sample and alternative elasticity checks give similar slopes.

Question 49

Results: Acemoglu & Restrepo (2022) - Tasks, Automation, and the Rise in U.S. Wage Inequality

Answer 49

Losing a large share of routine tasks (due to automation) is consistently followed by wage and hours displacement between industries (due to ripple effects). Automation explains much of the post-1980 widening of wage gaps and most of the rising college premium, even though it boosts overall productivity only modestly.

Question 50

Title: Cavounidis et al. (2024)

Answer 50

Obsolescence Rents: Teamsters, Truckers & Impending Computerization

Question 51

Motivation: Cavounidis et al. (2024) - Obsolescence Rents: Teamsters, Truckers & Impending Computerization

Answer 51

When workers know their occupation is likely to disappear (e.g. self-driving trucks), how do wages, hiring and workforce age adjust during the long lead-up? They label the temporary wage premium an obsolescence rent.

Question 52

Data / Setting: Cavounidis et al. (2024) - Obsolescence Rents: Teamsters, Truckers & Impending Computerization

Answer 52

1) Horse-drawn teamsters vs motor-truck drivers, 1910-30. 2) Modern U.S. occupations, 2005-19, matched to automation-risk scores. 3) Today’s long-haul truckers facing autonomous vehicles.

Question 53

Methodology: Cavounidis et al. (2024) - Obsolescence Rents: Teamsters, Truckers & Impending Computerization

Answer 53

Build a three-state overlapping-generations model (OLG) (no-shock / anticipatory-dread / aftermath), then test its predictions in the data. (1) compare wages, employment and age structure for teamsters in the dread decade (1910-20) versus the aftermath (1920-30); (2) Modern jobs ranked by Frey-Osbourne Automation risk (3) DiD, treatment: truck driver 2015-18 (dread), control: truck driver 2005-06 (pre-dread) - signs of workforce greying

Question 54

Robustness: Cavounidis et al. (2024) - Obsolescence Rents: Teamsters, Truckers & Impending Computerization

Answer 54

results are consistent for all three settings

Question 55

Results: Cavounidis et al. (2024) - Obsolescence Rents: Teamsters, Truckers & Impending Computerization

Answer 55

Rising wages and ageing workers in the occupation can be a warning sign that technology is on the way; labour-market adjustment (obsolence rent / wage premia) begins well before the machines actually arrive. 1) Teamsters show that wages first increase (dread stage) and then falls, with employment alllso dropping and median workforce age increasing 2) long haul drives are a contemporrary mirror of teamsters:

Question 56

Hydrostatics: Acemoglu & Restrepo (2022) - Tasks, Automation, and the Rise in U.S. Wage Inequality

Answer 56

metaphor for how automation reshapes task allocation and wages across worker groups. When automation displaces a group from its tasks, it creates ripple effects—those workers shift to other tasks, increasing competition and pushing down wages for similar groups.