PERI Living Wage Study

Comments of Dr. David Neumark
Visiting Fellow, Public Policy Institute of California
and Professor of Economics, Michigan State University

Ms. Susan McCarthy
City Manager
City of Santa Monica
1685 Main Street
Santa Monica, CA  90401

Dear Ms. McCarthy:

This letter contains my final comments on the report "Economic Analysis of Santa Monica Living Wage Proposal," by PERI. As you will be able to see from this letter and the earlier correspondence I have included, this review process has been a substantial task, involving a good deal of my time, as well as some research assistance. The undertaking has proven to involve far more than the two days work suggested by Professor Pollin originally. In part, this is attributable to the extreme length of the final report; the authors are to be commended for carrying out such a wide-ranging analysis. As a consequence, though, I have been forced to focus on a subset of issues, choosing those that I view as most pertinent to evaluating the effects of the living wage proposal.

In assessing my comments, it may be useful to you to understand where I stood on living wages prior to engaging in this process. My research on minimum wages is rather well-known. This research, which is based solely on empirical evidence, and not any insistence that economic models of minimum wage effects "must be right," points to negative effects of minimum wages, both in terms of reducing employment and hours of low-skilled workers, and increasing the proportion of families in poverty (as the employment and hours reductions of some workers in low-income families more than offset the wage gains introduced by the legislation). However, there is very little research on living wages. Consequently, I wrote the first paper (and so far the only one) that estimates the effects of living wages based on a comparison of experiences of cities that implemented living wages with cities that did not–using essentially the same methods as in the research on minimum wages. The paper discussed differences between living wages and minimum wages, and noted that because of more limited coverage, and because in some cases most of the added expense was likely to be borne by cities that could raise taxes to cover the expense, living wages could be less harmful or more helpful than minimum wages. Indeed, the research indicated that while living wages entail some employment losses, on net the evidence is most consistent with living wages reducing poverty rates slightly. (In Appendix 1 of the final report the PERI authors criticize this study. As you will see in my August 6 letter on the draft report, I think this criticism is misplaced or at least not fundamental to my study.) The bottom line is that I shouldn’t be automatically described as a "foe" of living wages; you can see the reference to my work in the September 4th  issue of Business Week.

Nonetheless, on a priori grounds I have some nagging questions about living wages as distinct from minimum wages. Prime among these is why who you work for (e.g., the city versus the private sector, or in Santa Monica’s case a Coastal Zone worker versus a worker across the street, or one with whose employer has more than $3 million in gross receipts, versus one whose employer does not) determines whether the government steps in and tries to raise your family’s income. I can imagine no rationale for drawing these distinctions among workers, which have nothing to do with their families’ needs. I’m also puzzled by why a smaller city would pass a law that essentially taxes employers located in the city to pay out benefits to many workers who do not live in the city. Finally, like minimum wages, living wages poorly target the low-income families we are trying to help, as the laws draw no distinction between the low-wage teenage worker of an affluent family and the low-wage single breadwinner in a poor family. For that reason, I am more supportive of income support programs, which target poor families well, can be directed toward residents of a particular jurisdiction, and if structured correctly can encourage work.

With that said, let me offer my review of the PERI report. I focus my comments on the strength of the evidence that the study’s authors report. Where it is relevant, I also discuss the earlier exchanges that occurred between me and the authors. Before plunging into my comments on the final report, it is probably useful to update you on the exchanges regarding this report that preceded the final report.

I initially reviewed the proposal, including the survey, in March of 2000. I have enclosed my March 16 letter to Professor Pollin, so you can see the complete content of that review. Many of the comments were minor, attempting to improve the survey the authors proposed using. In addition, I indicated that it was unclear how the authors were going to use the data they were collecting to draw conclusions about the likely effects of the living wage proposal. This is significant, because in the draft report that followed the survey data were used in a number of ways that hadn’t really been discussed before, and on which, therefore, I had been unable to comment.

Aside from these issues, in the March review I had one major criticism of the proposal. While the proposal’s stated aim was to assess how firms were likely to respond to the higher costs, the proposal never explained how this was going to be done. In particular, this proposed study made no attempt to infer what the Santa Monica living wage was likely to do, based on the experience with living wages in other cities, which is the standard way to evaluate policy effects in social science research. One counter-argument to this criticism is that the Santa Monica proposal is quite different from what other cities have done, so that past experience in other cities was unlikely to provide much guidance. As an alternative approach–probably inferior, but perhaps the best the researchers could do–I suggested that they augment the survey with questions posed to employers regarding their likely response to the living wage. Fortunately, the authors accepted this suggestion, and added such questions to their survey.

Subsequent to this, I reviewed the draft of the report in early August of 2000. I have also enclosed the full text of that review. As you will see, at that stage I had numerous criticisms. Here, I only touch on the main ones and those that have not been satisfactorily resolved. After explaining the initial criticism, I also offer my evaluation of how the authors have responded in the final report (and their letter to me accompanying the final report).

As noted above, the authors added to the employer survey questions regarding likely responses. Interestingly, for both restaurants and hotels a high fraction indicated they were likely to raise prices in response to the living wage. More importantly, the evidence pointed to a high likelihood of employment reductions in response to the living wages. For example, 40% of the restaurants said they were very likely to lay off workers, as did nearly half of the hotels, and 71% of the hotels said they would hire fewer workers in the future. To my surprise, though, the authors conclude, in both cases, that these results "send no clear message" regarding layoffs. As I indicated in my letter of August 6, I regard such conclusions as untenable. By way of analogy, I suggested that if Santa Monica were to pass a law forbidding restaurant employees from washing their hands, and as a result customers at 40% of restaurants got sick, while customers at 60% did not, we wouldn’t hesitate to draw the appropriate conclusions.

How have the authors responded to this? In his letter to me, Professor Pollin indicates "I am less inclined to project actual future behavior based on these responses. The problem is that the business respondents do not know–and cannot possibly know–the general equilibrium effects of this policy as it relates to them." This is a curious argument. By "general equilibrium effects," he means those effects that ultimately "shake out" after all price and wage increases have passed through all markets. It may well be true that the employers don’t understand these, but that’s a weak criticism. First of all, the partial equilibrium effects (i.e., those that we would predict based solely on responses in the single market of concern) may be sufficient to predict the effects of living wages. Second, the authors have no alternative framework for estimating the general equilibrium effects of a living wage increase. (There is a small set of researchers in economics who build general equilibrium models to evaluate policy effects, but none of that type of research is incorporated in this report.) Thus, I see no basis for believing that the authors of the report have a better idea of the general equilibrium effects than the respondents.

With that said, there is an argument against relying too heavily on these responses. Specifically, to the extent that the respondents believe their answers may affect the policy debate, and assuming they oppose the living wage, they have incentives to overstate the negative consequences of living wages. I have no way of knowing how important a consideration this is in this particular case. The bottom line, though, is that this evidence is still potentially among the most informative that the authors have, and it does point to adverse effects on employment (but not necessarily poverty).

In the final draft, the authors have taken this criticism seriously, and provided some estimates of potential job losses. (See, for example, Table 5.11.) The projected employment losses are not trivial. However, the authors have constructed arguments to try to trivialize these employment losses, which I think are not valid. First, on pp. 101-2, they correctly note that many of these jobs are likely to be of short duration, stating that "Most workers ... recognize that, in taking jobs with these firms, the probability is high that their job tenure will be brief, either through a voluntary quit or a layoff." This is irrelevant, however. It is true that low-skill jobs typically don’t last long, but the evidence the authors report does not imply that one worker’s employment will end, and another one will be hired with the same likelihood as before the living wage. Instead, it implies that there will be fewer low-skilled jobs available to these workers. That is, the employers’ responses must mean that positions would be eliminated in reaction to the mandated higher wage, not that worker A will be let go, and a similarly-skilled worker B will be hired.

The second way the authors try to trivialize the employment losses is to argue that they are a tiny fraction of the low-wage labor market in Los Angeles (p. 102). So what? They are an even tinier fraction of the labor force in all of California. The point is that the authors are not evaluating a living wage proposal for all of Los Angeles, and I assume that no one in Santa Monica who advocates a living wage thinks it will have an impact on the entire Los Angeles area. All that is relevant is what the net effect of the policy is on the affected workers, which is those who work or might work in the Coastal Zone. (This point also arose in comment number 4 of my August 6 letter.) The estimated employment losses ought to give one pause, because they could conceivably be large enough to more than offset the wage gains.

The preceding discussion covers the projection of employment losses based on the direct survey responses. Much of the report is devoted to projecting these effects in other ways, by using cost information to project how much living wages will raise costs for businesses in the Coastal Zone, and other information to project how this will affect prices charged by these businesses, with ensuing effects on demand for their products and hence ultimately on employment. An important part of the draft report argued strongly that the quantity of hotel rooms sold in Santa Monica was positively associated with price. As I pointed out in my August 6 letter, using this information to conclude that sales would not be negatively affected by price increases reflects a fundamental mistake in applied economic research, as shifting demand curves for hotel rooms may imply that the data are tracing out the supply curve (which should slope upward), rather than the demand curve (which should slope downward). This is important, because if demand doesn’t respond negatively to price, then employment will not be projected to fall in response to price increases induced by the living wage.

In their response letter, the authors fully acknowledge this error, and attempt to correct it using econometric techniques that allow one to estimate the slope of the demand curve, which is what they are after. This is explained on pp. 206-7. However, the authors have done this incorrectly. They include in the estimated equation variables that serve as proxies for income and wealth. But what they need to solve this problem is variables that shift the supply curve around (allowing them to trace out the demand curve), and to use these as "instrumental variables." But they neither have nor introduce any such variables. I’m afraid there’s really no way to explain this point without reference to technical issues, so you might have to take my word on this to some extent. But this component of the research is simply wrong and would not survive professional scrutiny.

Another major issue in the draft report concerned the statements made about the ages of workers covered by the living wage proposal. As you can see from my August 6 letter, my view was that the authors had made a rather serious error here, leading them to conclude that the proportion of teenagers among covered workers was conceivably lower than the proportion of teenagers in the workforce (which obviously couldn’t be the case). The authors have responded to my efforts to redo their calculations, but selectively. In particular, footnote 41 (p. 146) might suggest that I believe the 6.8 percent figure to be the best estimate. As my August 6 letter indicates, though, this is not the case, and I think higher numbers are more reliable. However, this is not a devastating point, as the fact remains that there are many non-teenagers among the affected workers. On the other hand, what is more relevant is not whether one might get a wage increase, but how big it would be. We know that at the lowest wage ranges, the representation of teenagers is much higher, whereas among those near but just below the proposed living wage, there are more older workers. Thus, any of these numbers severely understate the proportion of the income gains that are likely to accrue to teenagers, and conversely overstate the proportion of the gains that would accrue to older workers.

As I noted in my August 6 letter, I was troubled by some of the results on the share of family income earned by low-wage workers. Table 8.13 highlights the problem. This table shows that there are an average of 1.9 wage earners per family in the sampled population, but that the low-wage workers contribute well over one-half of family income, on average. This seems to me virtually impossible, since the other earners in the family must on average be higher-wage workers (assuming no workers below the minimum wage). I’m not quite sure what underlies this curious result, but the authors do not appear to have responded to this comment in the final report. I don’t doubt the basic conclusion that there are many low-income families among the potential beneficiaries of the living wage. But there are also many beneficiaries that are not in poor families, emphasizing the poor targeting of minimum wage types of laws. The authors have responded to this by including Table 8.8, showing that high proportions of the families of low-wage workers have incomes above nearly twice the poverty line. That’s not to say that these families couldn’t use more income; nonetheless, it is important to emphasize that this policy is not well-targeted at the poorest families.

Given what I communicated at the outset of this letter, I hope that my criticisms of this report will not be viewed as stemming from a desire to see the living wage proposal defeated. Frankly, I don’t have a strong view one way or the other. I doubt the living wage proposal will do a lot of good, but I also doubt it will do much harm. On the other hand, I have enough criticisms of this study to believe that it provides an insufficient basis to draw strong conclusions about the likely effects of the living wage proposal. In contrast, I think the authors often overreach, and draw conclusions that cannot be supported by the data and methods they use.

This page was last updated on 10/25/06.