The Compensating Wage Differential.docx

The Compensating Wage Differential The compensating wage differential is the additional income necessary to compensate for unpleasant working environment. The unpleasant working environment depends on individual preferences such as one person will need higher wage to accept a more dangerous job but others require less. Unpleasant characteristics can include increased danger, health risks, workload etc. The model: The labour market has workers with the same productivity The D is the disamenity of the job, if D is 1 then it it is dirty job and if D is 0 then it is clean job Preferences of the worker are given by their utility functions The utility depends on D and the wage of the job The w of an unemployed worker is 0 in the model

The Z is the compensation necessary to make the worker indifferent between the jobs The difference between the wages is the Z The Z is the reservation wage for job where D is 1 The workers have different preferences so the Z needed is different for each worker The indifference curves are different for each worker but are upward sloping

Market labour supply:

The worker preferences in the market are normally distributed If the w is greater than the Z given D for worker then they choose dirty job The w is the additional w the dirty job offers because of its worse conditions

The workers on the left of the w are choosing the job when D is 1 as the market wage is more than Z The workers on the right of w are choosing the job when D is 0 as the market wage is less than Z The workers have different preferences so there is a rising supply of labour for the dirty job as w increases The distribution is not always normal Firm strategy: Firms can use technology to clean job The cleaning of the job reduces D so the firm profits from offering a lower w to get workers The cost to the firm for the clean up is B If there is only job where D is 0 and where D is 1 then firm chooses between offering them If B is larger than w then firm chooses to give the job where D is 1 If B is less than w then firm chooses to give the job where D is 0

If B is decreased and is lower than than the w then the firm offers the job where D is 0 If w decreases and is lower than B then the firm offers the job where the D is 1

Market equilibrium and selection:

Market equilibrium is when demand of the workers is equal to the supply for each job The labour supply for jobs where D is 1 increases as w increases and the demand decreases There is selectivity bias in jobs where the D is 1 The workers that do the jobs where the D is 1 have the lowest disutility for the D The firms that offer the jobs where the D is 1 have the highest B The average worker that chooses the job where D is 1 is different from the average worker with D is 0 The systematic selection has important consequences for the inferences that can be drawn from the data

Measuring compensating wage differentials: The regression is the same as the Mincer regression with an explanatory variable for non wage benefits

The ft is the non wage benefits expressed as a percentage premium over wage The relation between the coefficient for the ft and the lnWt is expected to be negative There are problems using the regression because there are many unobservable individual charactarestics The individual preferences are unobservable so the Z is not known for each individual The percentage of earnings paid in non wage form is not known for the individuals Individual reference for types of jobs, bonus pay and the opportunities for promotion are unobservable Results are not always in favour of the compensating wage differential hypothesis, in particular for health insurance and paid vacation, the expected negative relationship is rarely established. Jobs that offer these non wages benefits are usually high paying jobs so the hypothesis is difficult to prove. Murphy and Topel show that higher unemployment risk is associated with higher wages. There is evidence for the compensating wage differential. The value of life: Using evidence on tradeoffs between risk and reward, economists have developed estimates of the value of a statistical life (VSL). The literature has concentrated on the observed wage premium and the size of the on the job risk. The existing literature based on estimates using the US labour market data typically show a VSL in the range of $4-9 million (Viscusi and Aldy, 2003). These VSL estimates provide governments with a reference point for assessing the benefits of risk reduction policies (e.g. limiting air pollution). Ashenfelter and Greenstone (2004) use a change in speed limits (from 55 mph to 65 mph) to estimate an upper bound on the public’s willingness to trade off wealth for a change in the probability of death. The trade-off was between a reduction in travel hours, therefore an increase in potential for production increasing wealth and an increase in fatalities. The adoption of the 65-mph limit increased speeds by approximately 4 percent, or 2.5 mph, and fatality rates by roughly 35 percent. These estimates suggest that about 125,000 hours were saved per lost life. When the time saved is valued at the average hourly wage, the estimates imply that adopting states were willing to accept risks that resulted in a savings of $1.54 million (1997 dollars) per fatality, with a sampling error roughly one-third this value. If the MB is larger than the MC then it will be socially efficient for the policy to be implemented. The MB is the hours saved per lost life multiplied by the average hourly wage in that area The MC is the potential earnings and other costs of the lost life

The discrimination:

There is discrimination when individuals with identical productive characteristics are paid differently because of non-productive characteristics. The discrimination is because of employer or worker preferences The discriminated workers have to accept a lower wage to compensate the employers In perfect competition there can be no discrimination The other firms will drive out discriminatory firms by hiring the minority workers They make more profits as they will pay lower than market wage but higher than the discriminatory firm The firms keep entering until then wage of the discriminated worker is equal to the market wage

Measuring discrimination: The difference in overall average income between majority and minority groups is societal economic discrimination. The difference in the average wage for equally productive majority and minority workers is labour market discrimination.

Wage regressions are used to measure wage discrimination It is difficult to determine an accurate estimation of the labour market discrimination The productivity of each worker needs to be controlled for but it is difficult to do it There is correlation between certain minority groups and productivity related characteristics Minorities have less education because of societal discrimination Women have lower work experience because of unequal job opportunities There are different tastes for workers e.g. some prefer market wage and others prefer non wage benefits Empirical analysis has problems but is useful as it helps to monitor discrimination over time in different contexts, it can be used to show policy makers what characteristics have large effects on earnings, it can also be used to determine whether a single firm is discriminating. It is difficult to distinguish whether the difference in wages are because of labour market discrimination or if it is statistical discrimination when the employers use characteristics such as race and gender to infer information about the workers that can make a difference in their productivity when other information is unavailable to them. Altonji and Pierret show that firms statistically discriminate among young workers using observable characteristics such as education as a signal for productivity. Then as firms learn about the true productivity, the coefficients on the easily observed variables fall as it becomes less useful as a signal. They find little evidence for statistical discrimination in wages because of race. Golden and Rouse (2000) used blind symphony orchestra auditions to conceal the candidate’s identity from the jury. They show that the screen increases the probability that a woman will be hired. The evidence is in favour of labour market discrimination. Bertrand and Mullainathan (2004) send fictitious resumes to help-wanted ads in newspapers with

randomly assigned African-American or White-sounding names. White names receive 50 percent more call backs for interviews. The evidence is in favour of either labour market discrimination or the labour discrimination. Some differences in wages exist as compensation for the non-monetary aspects of the job e.g. mortality risk, unemployment risk, non monetary benefits and local amenities. Empirical evidence is ambiguous. Murphy and Topel (1987) find that wages are higher if there is higher unemployment risk. There are many studies that find a positive correlation between wages and non-monetary benefits. The compensating wage theory can explain why discrimination between majority and minority groups of workers can arise due to preferences of employers and/or workers. Discrimination is difficult to measure as it shows wage differences for equally productive workers. Bertrand and Mullainathan (2004) use a field experiment to control for differences in productivity. There is some evidence for statistical discrimination.