Reading Summary Series: Trio on Job Market Polarization

The previous three papers established the observation of unequal wage distribution. ALM also established the discussion for computerization led SBTC as the cause for the structural wage change. These three smaller paper follow from previous discussion, and gave different view on the cause of wage change. They debate whether it is SBTC or polarization.

Autor, Katz and Kearney, “Measuring and Interpreting Trends in Economic Inequality: The Polarization of the US Labor Market.” AEA P&P 2006

This paper considers a revisionist view that wage market has a divergent trend in upper and lower tail inequality, i.e. polarization.  Data shows that upper half and lower half of wage distribution has diverging trend since 1987 for both male and female. There was faster wage growth in the bottom quartile than in the middle two quartiles, and the most rapid rise in the top quartile. This divergence also holds for residual inequality and robust for labor force composition change. According to ALM, routine tasks are complementary to high skill abstract task than to nonroutine manual tasks. Hence, computerization predicts that wage polarization is accompanied by employment polarization.

Building on ALM’s model, it shows that equilibrium occurs when economy operates on demand curve of the aggregate production function; each factor is paid at marginal product; and the labor market clears. From the model, manual and routine tasks are q-complements – rise in routine input raises marginal productivity of manual task input. It adjusts ALM’s model in that computer serves as a displacement to middle skilled routine tasks than to lowest skilled workers.

Autor, Katz and Kearney, “Trends in U.S. Wage Inequality: Revising the Revisionists.” RES 2008

This paper summarizes previous literature that two broad conclusions emerge for the wage and income inequality: 1. Much of the rise of earnings inequality in 1980s is due to shift in the supply of demand for skills combined with declining labor market institutions; 2. The surge of inequality in 1980s is due to secular rise in the demand for skill. However, recent studies challenged these conclusions in that: 1. Rise of inequality in 1980s is largely explained by nonmarket factors, such as declining real value of minimum wage; 2. Growth of earnings inequality was a one-time event of early 1980s. This paper re-evaluates these claims with special focus on: 1.growth of inequality an episodic event than secular phenomenon; 2. Nonmarket force of mechanical effect of labor force composition.

This paper uses March CPS and Outgoing Rotation Group CPS updating data from 1963 to 2005. It focuses on: 1. changes in overall wage inequality, captured by 90/10 percentile log wage differential; 2. changes in inequality in upper and lower halves of wage distribution, captured by 90/50, and 50/10 log wage gap; 3. Between-group wage differentials, captured b y college/high school wage premium; 4. Within-group wage inequality, captured by 90/10/, 90/50, 50/10 residual wage gap, controlling for education, age, experience and gender. Data shows a divergent growth of upper and lower tail wage inequality. And the expansion for between group differentials shows ups and downs.

Speculating the cause of expansion overall wage inequality, the paper suggests supply and demand factors, unemployment and minimum wage. It starts from examine education premia. CES technology model is imposed to test two hypotheses: skill neutral technological improvements or skill biased technical change. The Katz and Murphy model predicts growth of college wage premium through 1992. But after 1992, it over-predicts growth of college/high school premium. Along with subsequent models, it shows a slow down in demand that is inconsistent with SBTC. This paper explains it as low unemployment rate from cyclical labor market expansion. Data since 1987 show a polarization trend in earnings. This makes observations inconsistent with two-factor CES model. In looking at experience by education, simple supple and demand framework focuses a secular increase in relative demand for college workers combined with fluctuations in relative skill supplies. This paper points that SBTC has the main role: 1. Relative employment of more educated workers and nonproduction workers has increased within detailed industries; 2. Adoption of computer based technologies and the increased use of college educated labor within each industries.

In arguing the role of minimum wage, this paper finds that it is not important in the change of educational wage differentials, since trends in between-group inequality and trends in overall residual inequality differ. Residual inequality stabilizes since 1990, while overall inequality still grows. Data shows that minimum wage correlates to both low tail inequality and high tail inequality. This shows spurious causation.

The paper also finds that labor force composition change does not explain diverging path of upper and lower tail inequality. It further shows that demand forces played a key role in shaping wage structure. These patterns is explained by richer version of SBTC hypothesis in which technology complements highly educated workers and substitutes for moderately educated workers, and has less impact on low-skilled workers performing manual tasks.

Goos and Manning, “Lousy and Lovely Jobs: The Rising Polarization of Work in Britain.” RES 2007

This paper follows from ALM’s structure and focuses on the argument of explaining wage inequality data to be SBTC or polarization. It argues that the impact of technology will be to lead to rising relative demand in well-paid skilled jobs and in low-paid least skilled jobs, and failing realative demand in the middling jobs. This is referred to as job polarization. It shows that an increase in the relative demand for low wage workers is not in line with the SBTC hypothesis, since SBTC hypothesis only indicates the increase in high skill jobs and all the rest should decline.

This paper copies ALM’s job coding from DOT, and focuses on the angel: jobs that can be routinized are not distributed uniformly across wage distribution. ALM argues that nonroutine cognitive and interactive tasks are complementary to technology, and routine tasks are substitutes, and nonroutine tasks are not affected. But this paper argues that general equilibrium effect indicates that employment first shift towards jobs in low productivity growth sectors to keep balance of output in different product.

Data shows that there are large growth in the share of employment in top two deciles, smaller growth in share of jobs in bottom deciles, and sharp decline in middling jobs over the time period from 1979-1999. The J shape relationship does not support SBTC hypothesis. Regression estimates also provide quadratic term between employment growth and initial wage.

It also shows that top growing job focus in finance and business sectors that also locates at the top wage distribution. High in the distribution of employment also includes care, education, and hospital assistants, as low paid jobs. The paper infers that job polarization could be drive by factors other than technology, such as changes in labor supply or changes in demand other than technology.

Changes in labor supply can be from: 1. Increase in feminization of labor force; 2. Increase in education level; 3. Age structure change; 4. Immigration. However, counterfactual exercise shows that supply does not help in explaining growth in lousy jobs; though it does help growth in lovely jobs. Demand change not from technology also has: trade and structure of product demand. Unfortunately, none of the hypotheses have the ability to explain job polarization.

Further analysis show that ALM’s routinization hypothesis still generally stands for the polarization phenomena. Lower tail wage inequality decreases as the scarcity of displaced workers drives up the relative wage of works in lousy jobs.

Reading Summary Series: Autor, Levy and Murnane, “The Skill Content of Recent Technological Change: An Empirical Exploration.” QJE 2003

Autor, Levy and Murnane, “The Skill Content of Recent Technological Change: An Empirical Exploration.” QJE 2003

This is one of my favorite paper in the topic series of wage inequality distribution. It not only has creative ideas in gauging and measuring task components in each industry and occupation, but also has a very fine general equilibrium construction, which leads to a very logically cohesive development from theory to empirics. The intuition explanation from the model also shines bright. Overall, it also leads to a new branch in the discussion of skill biased technical change.

Previously discussions of SBTC focuses on drawing the correlation of computerization and labor demand change. This paper focuses on causal inference. It observes that: 1. computer capital substitutes for workers with limited and well-defined set of cognitive and manual activities, i.e. routine tasks; 2. It complements workers with problem-solving and complex communication tasks, i.e. nonroutine tasks. It shows that initially routine task intense industries will make relatively large investment in computer capital as its price declines, which reduces routine task workers and increases nonroutine task input, as represented by highly educated workers with comparative advantage in nonroutine versus routine tasks.  Greater intensity of routine inputs increases marginal productivity of nonroutine inputs. Based on observations, the paper further splits routine and nonroutine tasks into manual tasks and analytic and interactive tasks.

The paper assumes CRS Cobb-Douglass production function with routine and nonroutine labor inputs and computer capital as production factors. Computer capital is perfectly elastic at market price, which is further exogenously determined. Assumption has perfect substitutability between computer capital and routine task labor. And these are all relative complements to nonroutine labor. Workers have heterogeneous endowment in routine and nonroutine tasks. Therefore, workers choose tasks composition based on comparative advantage. Market clearing conditions have that: 1. wage of routine tasks equal to price of computer capital; 2. Worker routine and nonroutine tasks clear labor market. From the model, we can see that decline in computer price reduces wage for routine tasks 1-1, and raises demand for routine task inputs, which further raises nonroutine tasks wage. Hence, marginal worker reallocate labor input from routine to nonroutine tasks, and demand increase for routine task is fulfilled by computer capital. From the model, the causal force is computer capital price.

Further detailing the model into industry level, it assumes all industry Cobb-Douglas technology, First order conditions return profit maxed wage rate and factor demand function. From these derivations, it proposes that: 1. each individual firm’s degree of adoption to computer capital depends on industry specific factor share of nonroutine tasks, despite the same computer price; 2. Decline in computer price raises demand for nonroutine task input, and the scale is larger in routine-task-intensive industries; 3. Larger investment in computer capital show larger increase in nonroutine labor input and larger decrease in routine labor input.

Empirically, the paper uses Dictionary of Occupational Titles to decode industry task composition. IPUMS and CPS Merged Outgoing Rotation Group are also used. Job task change over time is measured by: 1. Extensive margin: Occupational distribution of employment, holding constant task content within occupations at 1977 DOT level; 2. Intensive margin: change in task content measures within occupations over the period 1977 to 1991. The paper further transform DOT measure of tasks into percentile values corresponding to rank in 1960 distribution of task input.

To test the first proposal, it hypothesizes that historically routine task intensive industries should adopt computer capital more rapidly as its price fell. Simple predictive regression and robustness check confirm the hypothesis.

To test for the second proposal, it hypothesizes that declining computer price should reduce aggregate demand for routine task labor and increase for nonroutine task labor. By pairing task measures with employment data for each decade, it shows that share of labor force employment in nonroutine cognitive and analytical intensive task increased substantially, and the opposite for routine tasks.  Secular decline is also found in nonroutine manual tasks. It shows that distribution of nonroutine analytic and nonroutine interactive task input grew substantially. And the routine cognitive and routine manual decreased substantially. When further split data by gender, both genders show similar trends; but female has numerically larger shift. When decompose the shift by extensive and intensive margin over the 20 year interval, both nonroutine analytic and nonroutine interactive task measures show strong and accelerating within-industry growth. Routine cognitive and routine manual acceleratingly declined since 1970s due to within industry shift. In summary, trends dominating the change is from within-industry shift.

Connecting these shift back to computerization, regression of change in computer on change in input task is conducted for each decade and for within industry and between industry. Results show that almost the entire observed within industry change in nonroutine task change is explained by computerization, so is for between industry changes. To further test the robustness of these conclusions, the paper also uses principal component analysis and computer capital investment data from National Income and Product Accounts. All confirms the original conclusion, and confirms the third proposal about larger computer capital investment.

Further, the paper claims that change in demand for workplace, from technological change, is the underlying cause of relative demand shifts favoring educated labor. As previously shown, computerization adaptation predicts increase in nonroutine workers and decrease in routine workers. The paper conjectures the reasons due to: 1. industry purchase computer needs better educated workers to master these tasks; 2. Industries change task assignment of workers with given educational attainment to more of a routine task assignment. Regression within group is estimated. However, holding computer adoption fixed, the model cannot predict within industry task change for high school and college group. The paper explains that the reason is due to “topping out” for college graduates, since they are already at the top of distribution, and high school dropouts have too less human capital to effectively adopt to new technology requirement.

To quantify the task shift, the paper draws task changes within industry, education groups, and occupations to calculate demand for college educated workers. With imputed elasticity of substitution between high skilled and low skilled labor inputs, it finds that extensive margin task change explains 20% to 25% of estimated demand shift for college versus noncollege labor. 40% of computer contribution to rising educational demand in the last two decades in sample period is due to shift in task composition within occupations.

Reading Summary Series: Juhn, Murphy and Pierce, “Wage Inequality and the Rise in Returns to Skill.” JPE 1993

Juhn, Murphy and Pierce, “Wage Inequality and the Rise in Returns to Skill.” JPE 1993

Last paper (K&M, QJE 1992) serves as an overture of modern labor market inequality research. One small niche tangent to this paper was that K&M used residual distribution to look at within group demand shift. Another niche related to this one is in the section of analyzing education wage premium, K&M listed two explanations: different premium associated with skills (return to education), and different skills from varying levels of education.

This time, Juhn, Murphy and Pierce look at trend in inequality, through which they focus on unobserved factors that affect inequality. They argue that inequality is about distribution of skill and skill price. The highlight of this paper is the decomposition of wage distribution into distribution of observed skill, change in price of skill, static distribution of residuals, and change in residual inequality.

This paper uses March CPS data from 1964-1990 and 1960 decennial census. From data, during 1963 – 1989, there is an overall wage increase. However, the increase is mainly from the top percentile increase. The middle percentile and the lower percentile wage experienced certain level of decrease. This divergence also showed in education level and experience level. The paper interpret it a divergence between the most skilled and the least skilled. The paper chose wage as a measurement due to its direct reflection to market price. It also decomposes increase in inequality into observed qualities and differences within these qualities. The dispersion in wage after controlling for observables, is interpreted as distribution of unobservables valued in the market. From observing timing of change in wage structure, the paper concludes that rise in inequality and rise in education premia are different economic phenomena. This paper finds that rise in return to skill is because of skill biased technology shift. The paper also finds that changes in composition of work force and changes in pattern of employment across sectors affect level of wage inequality, but unimportant.

The paper decompose real weekly wage distribution and look at 10^th percentile, median and 90^th percentile wage. Median wage increased from 1963 to 1973, but having ups and downs till 1985. The 10^th percentile wage rose from 1963 to 1970 by 20%; then declined by 25% till 1989. But real wage at the 90^th percentile, although declined from 1973 to 1975, has been increasing steadily and gained about 40% in 1989 compared to 1963. Real hourly wage shows the same trends but more volatile. In general, inequality has increased substantially in all parts of wage distribution. When decompose the wage into years of experience group, the inequality has increased between experience group, as well as within experience group, with the lowest groups of young workers having lower real wage in 1989 than in 1963. 2/5 of all young workers have no change in real wage at all during the period. Similar is the story if decompose workers into college graduates and high school graduates.

Controlling for education and experience, residual shows stability from 1959 to 1970, and shows enormous change in dispersion from 1970 to 1988. The increase in residual represents within group wage inequality, which is interpreted as a trend towards higher skill price. It can also be the increase in education inequality. Inequality within cohort is examined to verify the second claim. Within cohort change is resulted from age/cohort and time effect. From analyzing cohort changes, it confirms that the rise in inequality is due to rise in skilled price, not because of rising cohort quality.

As indicated in the beginning, this paper’s main contribution is to decompose change in wage inequality focusing on residual dispersion. In this paper, change in inequality comes from A. Change in individual quality; B. change in price of observables; C. change in distribution of residuals.  From data, changes in observed quantities only account for about 7% of the increase in wage inequality below the median and 14% of the inequality above the median. The rise in observable price accounts for about 47% increase in above median inequality and only 26% below median. Therefore, unobserved qualities accounts for 65% of increase in inequality below the median, and 39% of the increase in inequality above the median. Based on these findings, it infers this change due to return to skill within education and experience group, not cross groups.

Large shift in skill premia is inferred to be from demand shift towards most skilled, either across industries or within industries toward production methods favoring more skilled. As a result, shift in industrial and occupational composition across sectors generate significant changes in relative demand.

According to the model and demand index, rising demand for skill has characterized the period. The demand for skill increased but the skill premia did not change. But it does not explain the contrast between events of the 1960s and the last two decades. The change in occupation and industry composition suggests that wage inequality rose as a result of shift in employment towards high wage jobs, as supported by the demand index. The paper decomposes variance of wage change into variance of across industry and variance within industry. Composition effect raises within industry variance. Skill price effect affects between industry variance.  Therefore, along with dada, fall in wages for the least skilled is due to fall in demand for low wage workers, rather than rise in number of low skilled jobs.

In summary, both supply and demand grew for highly educated workers. But demand growth more significantly than supply growth, which results in the wage premia. The increase in return to experience, however, can be attributed to the arrival of baby boomer and the associated youthening of the labor force.

Reading Summary Series: Katz and Murphy, “Changes in Relative Wages: 1963-1987: Supply and Demand Factors.” QJE 1992

It has been a few weeks since the new school year started, and I have accumulated a lot of readings. Here I want to make a summary of the ones that I think are really good.

Katz and Murphy, “Changes in Relative Wages: 1963-1987: Supply and Demand Factors.” QJE 1992

This is about one of the first as well as the most important modern paper comprehensively looking at the inequality issues in the US from a demand perspective, using a supply and demand framework. It is quite a long paper and many of the things it talks about seem a bit cumbersome from the current standpoint. But it is the forerunner, after all.

This paper takes a look at the US wage structure from 1963 to 1987. Data shows that there were rapid growth in wage premium for more-educated workers, more skilled workers, and females. This paper concludes that the main cause is the change in relative demand in the labor market. A few variants are: 1. Technological change; 2. Shifts in product demand associated with large trade deficit in 1980s that led to decline in manufacturing employment; 3. Changes in wage setting institutions, such as decline of union and the decline in real value of minimum wage.

The paper uses 25 consecutive years of data from CPS March Supplement. It created 320 distinct labor groups by sex, education, and 40 single-year potential experiences. The wage measure is average weekly wage of full time workers. The counter sample measures total hours worked for each group by aggregating the product of total annual hours and individual CPS sample weight.

In cross group comparison, over the period, average wage increased by 16.1%. Average wage for women increased by 9% relative to men. This narrows wage gap between male and female, especially in the 1980s. The trend of narrowing gap started earlier for college graduates, but more substantial for high school graduates. In terms of education level, real wage monotonically increased by education level. But it has interesting up and downs between college and high school graduates. From 1963-1971, college graduates gained, while between 1971 – 1979, college graduates wage fell, and between 1979-1987, it gained exponentially. The variation of wage differentials by education is larger for young workers than for older workers. In terms of experience, wage differentials between young and old grew larger over the entire period. These cross group comparison only accounts for 1/3 of the wage differentials across workers. The more is accounted by within group changes.

Within group change is examined through distribution of residuals from regression of log weekly wage on demographic group controls. Residual distribution captures dispersion of wage within the demographic groups. It shows that within group inequality has expended largely for both women and men. Expansion of cross group differentials shows that less-educated and less-experienced workers lost out over the period; expansion of within group differentials shows that “least-skilled” workers within each group lost out. Data shows that overall residual for both men and women was stable in 1960s, then increased by 30% from 1970 to 1987.

In a supply and demand framework, the paper views relative wages of different groups generated by relative supply of the group and the factor demand from aggregate production. Due to the concavity of aggregate production function, changes in factor supplies and changes in wage must negatively covary. That is to say, if the factor demand is stable, wage change and labor input change must be negatively correlated.

In looking at relative supply, average educational attainment of labor force has the most prominent change. While college graduates’ labor force almost doubled, high school dropout’s fell about 2/3. Comparing from the same positive wage change, it infers that relative demand has shifted.  It also shows that largest increase in supply of college graduates is during 1971-1979. And that’s when college wage premium declined. While as the next decade, supply drops and wage premium increased. That explains the trend in college wage premium.

As for women, their labor force growth increased rapidly in 1970s than in 1960s and 1980s. With secular growth in relative demand for women concentrated industries, this explains the wage growth for women in 1980s than in 1970s. Similar stories exist for workers with experience versus younger workers.

When formally examine causal relationship between relative supply change and relative wage change, it shows that groups with largest supply increase also have the largest wage growth. This infers the importance of demand change.

The paper further analyze demand shift. It decomposes demand shift into within industry shifts and between industry shifts. Within industry shifts is about changes of relative factor intensities within industries at fixed relative wage, while as between shifts is about changes in allocation of total labor demand across industries at fixed wage. Between industry demand shift depends on group differences of employment distribution. The shift over the whole period shows the outflow of low tech and basic manufacturing and inflow of professional and business services, which favors college graduates and women.

Suppose there is only between industry demand shift and relative supply change, all that has positive demand shift with increase of wage should have decreased share within each industry. But data doesn’t agree. This shows within industry demand shift. Similar to the overall demand shifting model is constructed for between and within industry demand shift.  Measurement is constructed using industry*occupation categories. Between industry shifts for college graduates decelerated in the 1980s, and within industry demand shifts have accelerated.

This paper further examines demand shifts resulted from international trade. Equal allocation method treats net imports as an analogous to domestic production for domestic consumption. Production worker allocation method treats exports allocated to all in the, but imports allocated to production workers only. Data shows that effects on relative labor demand of trade were moderate until substantial trade deficits in 1980s. This negatively affects high school dropouts, especially female dropouts. It tends to explain the education wage premium differentials. But, the relative supply change is more dramatic than the impact of international trade.

To understand the education wage differentials, several interpretations have: 1. Changes in relative earnings represents changes in relative market price of skills by college and high school graduates. 2. Composition of college and high school graduates affect the relative skill level.

CES production function model depicts the relationship between high school and college school graduates in terms of labor input factors. The elasticity of substitution between college and high school equivalents shows the change in relative demand given fixed relative supply and price. The greater the elasticity, the higher change in quantity when wage changes. Low elasticity of substitution indicates smooth change in quantity when wage changes.  OLS and imputation methods are used to approximate the value of the elasticity. 1.41 is approximately the reasonable value for the elasticity of substitution between college and high school labor.

Lastly, the paper examines experience differentials. Pattern of changes in experience differentials for all men is dominated by changes for those with less than 16 years of schooling. This is an evidence for the “active labor market” hypothesis. It tells that changes in labor market show up most sharply for new entrants, because senior workers are insulated by labor market institutions. The shrinking of manufacturing sectors in the 1980s is impacting on young less educated males from such hypothesis.