Question

The authors include Table II in the paper (on the next page) to demonstrate that there is a strong relationship between computer use on the job and wages in both the United States and Germany, and that the relationship has grown stronger over time. Computer use is actually more strongly related to wages than years of schooling or experience.

(a) When considering whether this relationship is causal, what is an omitted variable that could potentially be related to both on-the-job computer use and wages?

(b) Does that omitted variable in part (a) lead to overstating or understating the “effect” of computer use on wages?

2. Table III from the paper (on the next page) presents the results of several regressions of wages, the dependent variable, on various tools used on the job, the independent variables. In Germany in 1991-92, not including any occupation controls, we see that using a computer on the job was associated with 17% higher wages while using a pen or pencil was associated with nearly 13% higher wages. Now focus on the models for Germany in 1991-92 including occupation controls.

(a) What is the coefficient on computer use? What is the coefficient on pen or pencil use?

(b) Comparing this column of results to column (3) discussed above, why does the inclusion of controls for occupation make such a big difference in the estimated coefficients?

Returns to Computer Use In a famous paper in the Quarterly Journal of Economics (1997), DiNardo and Pischke explore the returns to on-the-job computer use, i.e., do those who use computers on the job earn more? Previous work had suggested yes, and that these earnings differentials captured the labor market returns to computer skill or knowledge. Here is the paper abstract: THE RETURNS TO COMPUTER USE REVISITED: HAVE PENCILS CHANGED THE WAGE STRUCTURE TOO?* JOHN E. DINARDO AND JÖRN-STEFFEN PISCHKE Are the large measured wage differentials for on-the-job computer use a true return to computer skills, or do they just reflect that higher wage workers use computers on their jobs? We examine this issue with three large cross-sectional surveys from Germany. First, we confirm that the estimated wage differential associated with computer use in Germany is very similar to the U.S. differential. Second, we also measure large differentials for on-the-job use of calculators, tele- phones, pens or pencils, or for those who work while sitting down. We argue that these findings cast some doubt on the literal interpretation of the computer use wage differential as reflecting true returns to computer use or skill. 3 Interpreting the results. 1. The authors include Table II in the paper on the next page) to demonstrate that there is a strong relationship between computer use on the job and wages in both the United States and Germany, and that the relationship has grown stronger over time. Computer use is actually more strongly related to wages than years of schooling or experience. (a) When considering whether this relationship is causal, what is an omitted variable that could potentially be related to both on-the-job computer use and wages? (b) Does that omitted variable in part (a) lead to overstating or understating the "effect" of computer use on wages? 2. Table III from the paper on the next page) presents the results of several regressions of wages, the dependent variable, on various tools used on the job, the independent variables. In Germany in 1991-92, not including any occupation controls, we see that using a computer on the job was associated with 17% higher wages while using a pen or pencil was associated with nearly 13% higher wages. Now focus on the models for Germany in 1991-92 including occupation controls. (a) What is the coefficient on computer use? What is the coefficient on pen or pencil use? (b) Comparing this column of results to column (3) discussed above, why does the inclusion of controls for occupation make such a big difference in the estimated coefficients?

Answer 1

1.a>

The omitted variable that potentially is related to both on the job computer -use and wages is education. (here years of schooling. The higher the years of schooling a person has, the more wage he earns and he is more likely to use a computer.

b>

Yes, this omitted variable (years of schooling will overstate the effect of computer use on the wage. As we see in column table 2., for the US in 1984, with one additional use of computer will lead to increase in wage by 17% percentage points while when we add years of schooling as a covariate, we can see with the additional use of the computer, the wage gets increased by 6.8 %. This is because the rise in wages might be partially due to attaining schools.

2.

table 3. For Germany 1991-92, On an average, the additional effect of one extra use of a computer on wage is 8.3% points while  On an average, the additional effect of one extra use of a pen or pencil on wage is 5% points.

Coefficient of computer use -0.083,

coefficient on pen or pencil - 0.050

b>inclusion for control for occupation makes a big difference in the estimated coefficient because we have added the controls for occupation so that the wage difference which is attributed due to changes in occupation structure gets eliminated. when we add controls, the estimated coefficients gets smaller as they truly reflect the effect of the use of a computer on wage and eliminates the fact that some occupation like a white-collar job needs high use of computer and thus have a huge effect on wage