In their paper "Exploring the Complaints and Compliance Gap under U.S. Workplace Policies" David Weil and Amanda Pyles argue convincingly that if the number of worker complaints is found to be correlated with workplace violations, then a workplace regulatory system that is reliant on complaints might be said to functioning properly. Using Occupational Safety and Health Administration (OSHA) and Fair Labor Standards Act (FLSA) data at the industry level, they then examine the relationship between complaint rates and measures of workplace violations for several recent years. They find the following:

¥   That complaint rates are low relative to the number of violations

¥   That there is a high level of variation in complaint rates across industries

¥   That the correlation between violations and complaint rates is low

     Overall, the paper is nicely done and well argued. I have three concerns about their data, however, and also a few suggestions concerning their methodology. With respect to data concerns, their FLSA noncompliance variable is measured only by overtime violations, and not also by minimum wage violations. It is not completely clear why minimum wage violations are not also included, although the authors do state that the "vast majority" of wage and hour law violations are of the overtime variety. The authors also state that they have relied on worker responses concerning industry affiliation rather than on admittedly more reliable employer responses. Finally, their complaint rate data for OSHA violations do not include cases where the worker complaints were resolved by telephone. In all three cases, it would be nice to have a clearer idea as to why the variables were so specified and also some indication of the possible magnitude of the consequent error.

     Concerning their methodology, the authors first compare the top ten and then the bottom ten industries in each category in order to see how much "overlap" exists. In other words, do those industries with the highest noncompliance rates also tend to have the highest complaint rates (and vice versa)? They find that not very much overlap does exist—especially among the top ten in each of the two categories. The comparison is an interesting one, but a much better way of describing the degree of association across all the industries would be the Pearson coefficient of correlation or the Spearman rank correlation coefficient.

     On a related note, the OSHA regressions that the authors estimated show what I would call reasonably high R²values of between .25 and .30; and these R² values would in turn imply (if the regression equations were bivariate, which they are not) Pearson correlation coefficients of between about .50 and .55. For cross-section data with possible measurement problems of the type that I discussed previously and with such a parsimonious model, these R² values signify a pretty good fit of the regression equation to the data. The authors downplay their results somewhat, however, in saying that workplace violations explain "relatively little" of the volume of complaints. (Of course, the same thing cannot be said of the overtime violations regressions, whose R² values are very close to zero.)

     One last suggestion: the authors might consider testing for endogeneity. For example, the OSHA complaint rate may well influence the OSHA compliance rate, as well as the other way around. All in all, though, my observations and suggestions should not be taken to detract from what is a most insightful and valuable—especially from a policy perspective—paper.

     The paper by Kristen Monaco, "Wages and Working Conditions of Truck Drivers at the Port of Long Beach," gives us a very nice case description of port drayage driver employment. As she states—and here is the real contribution of her paper—"There is very little known about these drivers [other than] anecdotal evidence." Furthermore, ever since the events of 9/11, security at U.S. ports has become a national concern. All of us remember the flap just last year about the proposal that would have let a United Arab EmiratesÐbased company run some key U.S. seaports. Thus, knowing more about this particular labor force is important. In addition to her descriptive statistical portrait of these drivers, their pay, hours, ethnic background, education, and employers, she has also presented some regression estimates of the determinants of the earnings of these drivers, their wait times for loads, and their willingness to accept unsafe conditions.

     I will not go into detail about her regression results, which are adequately summarized in her paper; but I would like to offer a couple of suggestions and raise a few questions concerning some things that I found unclear. First, she states that the "self-employed are not allowed to form a union under current anti-trust laws." I believe that a more accurate statement would be that the self-employed are not guaranteed rights to join unions under current labor laws. And in the very next line Professor Monaco mentions that "the Teamsters currently have a focused campaign to organize port drivers." I think readers would be interested in knowing a little more about the Teamsters' efforts in this area.

     I also detected a couple of statistical discrepancies that should be cleared up. She states that the majority of drivers in her sample are owner-operators; however two different percentages are given in different places in the paper concerning the number of owner-operators in the sample. In addition, in the text mean net income of drivers in her sample is reported as $29,903, while in the appendix it is stated as $35,436.

     I also have one question concerning the specification of her "wait time" variable. As she states, a key issue facing drivers is the amount of time they spend waiting at the ports for their loads. In analyzing the determinants of driver waiting time, she uses as the dependent variable the ratio of waiting time to total time of the last trip. However, if there happens to be a lot of variation in wait times across trips for individual drivers (and I suspect that there would be), then this variable is less than satisfactory. Because the information that she has collected is from a self-administered questionnaire to drivers waiting for the gates to open, it may have been the case that this information is the best that she could reasonably expect to be reported.

     The comments above notwithstanding, this is a very informative paper. And for the benefit of other researchers wishing to perhaps do similar surveys of their own, I would suggest that she add a copy of her survey questionnaire to the appendix of her paper.