Discussion Paper

No. 2009-35 | July 27, 2009
Selection Wages and Discrimination


Applicants for any given job are more or less suited to fill it, and the firm will select the best among them. Increasing the wage offer attracts more applicants and makes it possible to raise the hiring standard and improve the productivity of the staff. Wages that optimize on the trade-off between the wage level and the productivity of the workforce are known as selection wages. As men react more strongly to wage differ¬entials than females, the trade-off is more pronounced for men and a profitmaximizing firm will offer a higher wage for men than for women in equilibrium.

Data Set

Data sets for articles published in "Economics" are available at Dataverse. Please have a look at our repository.

The data set for this article can be found at: http://hdl.handle.net/1902.1/14248

JEL Classification

B54 D13 D42 J31 J7

Cite As

Ekkehart Schlicht (2009). Selection Wages and Discrimination. Economics Discussion Papers, No 2009-35, Kiel Institute for the World Economy. http://www.economics-ejournal.org/economics/discussionpapers/2009-35


Comments and Questions

Anonymous - gender discrimination
July 28, 2009 - 12:37

I think this is a very sophisticated paper in terms of mathematical and econometric elaboration. Nonetheless, I think it lacks more theoretical reading regarding the reasons behind gender wage discrimination. In my view the relevance of the argument can only be tested including other omitted variables, related to the family ...[more]

... condition of these women, human capital, marital status. Therefore in my opinion it is better not to recommend the paper for publication now but to encourage the author to enlarge the scope of the analysis.

Anonymous - Reader Comment
July 29, 2009 - 09:10

See attached file

Ekkehart Schlicht - Social Categories
August 03, 2009 - 13:00

Thank you for your valuable comments. I have to think a little longer about some of them, but maybe I can clarify a few points right now.

But first on the paper: It was intended as a theoretical paper pointing out a possible mechanism for discrimination, without excluding others. ...[more]

... My rudimentary discussion of existing approaches was intended to clarify that I do not see the idea outlined here as excluding other mechanisms. It is not a full-fedged theory of discrimination (which would require a book, I presume). You have just a note applying the selection wage idea to the issue of discrimination.

I see "family considerations" and an emphasis on “non-monetary job attributes" related to each other--as complementary, rather than exclusive.

The commentator raises, among other issues, an intereasting point that deserves close scrutinity:

"the model does not assume two different groups of women, one with strong 'family considerations' (more responsive to non-monetary job attributes) and the other with weak 'family considerations' (more responsive to monetary attributes) and does not demonstrate how the selection mechanism would work."

I would defend the position taken in the note by proposing that women and men can be readily distinguished by an employer, whereas family women and work women cannot. Hence no signals are available that distinguish female applicants according to type; or the type information can be easily hidden or faked, rendering it useless for the employer. Discrimination cannot build on that.

It may be argued against that position that we had (and sometimes have) discrimination according to religious affiliation, were the same argument would seem to apply. Yet this type of discrimination has disappeared, along with a visibility of religious affiliation. It may re-appear if it is becoming more visible, costly to fake, and related to economically relevant differences, but this would be quite in line with my argument. But maybe there are other relevant cases.

The commentator's remark could, however, be read differently. I could suggest that the male/famale distinction is socially constructed; and, without doubt, it is. We may easily conceive a world where social class or caste affiliation is more important than sex, for instance, and where discrimination occurs mainly according to that affiliation. Hence a theory of discrimination builds ultimately on socially constructed categories that derive from the laws that govern our categorizations. We economists may take them as given, but we should not fall into the trap of thinking that they are arbitrary. Further we should be aware that such categorizations induce self-categorizations (identities) that directly affect behavior. This renders them economically relevant. The psychological laws governing concept formation, stereotyping, etc. generate concepts and categories that capture the world around us as it is, and these concepts and categories (such as male and female) affect economic processes in turn. I take the remarks of the commentator as a reminder that we have to think much deeper than we usually do when penning down economic models.

I have a problem to understand what alternative kind of discrimination theory is conceivable, or what the commentator has in mind here. This would be interesting.

The suggestion to apply the argument to different social strata is a very valuable one, as it may help to identify reasons for discrimination. I shall think about that. The most interesting, but a little lower-key, application would be the urban/rural dimension, where the view outlined in the paper carries implications. A strict application in Germany is problematic, as discrimination is illegal.

Finally, the commentator has been right in detecting my uneasiness with the point that different sensitivities of women may induce parrallel mechanism relating to non-monetary incentives. I have skipped that for the sake of the argument, but I concede that I find it difficult to handle the issue in a meaningful way.

Anonymous - Reader comment on "social categories"
August 04, 2009 - 17:05

August 3rd, 2009
The author aptly reminds the limits of his paper and concedes the need for further investigation of different aspects of his discriminatory mechanism in the future. There remains, however, a controversial issue concerning the possibility of modeling the discriminatory mechanism with two groups of women, namely family ...[more]

... women and working women.
The author claims “no signals are available that distinguish female applicants according to type; or the type of information can be easily hidden or faked, rendering it useless for the employer. Discrimination cannot build on that.”
If I follow the author’s line of argument, discrimination exists since there is no mechanism to sort out working women from family women. In this case, discrimination does not originate from women’s preferences but from the lack of a proper signaling mechanism (or due to a very costly signaling mechanism) to sort out women’s with different private preferences. But if so, the result would be inconsistent with the basic argument of the paper according to which discrimination is caused by women’s preferences.
In Akerlof and Kranton’s seminal paper (2000), women’s utility functions are defined in a way to capture their gender identity “payoffs” derived from their own actions. But in Schlicht’s discriminatory mechanism, what causes discrimination is not identity based payoffs of all women’s utility function. For even with women having the same “payoffs” or the same responsiveness to monetary attributes of job as men, discrimination will continue as long as there is no readily signal to sort them out from other types of women. It seems to me that the discussion about the assumption of two different groups of women was particularly fruitful to scrutinize the underlying discriminatory mechanism in Schlicht’s paper. However, his reply does not bring more clarity in this respect despite the great qualities of the paper as mentioned in my previous comments.

Akerlof G. and Kranton R. (2000), “Economics and Identity”, Quarterly Journal of Economics, Vol. 115, No. 3, pp. 715-753.

Claus Schnabel - Stimulating paper
August 05, 2009 - 14:10

I think this is a very interesting and stimulating paper. As far as I can see, it is mathematically sound and provides some new insights. What I am not sure about yet is how to distinguish empirically between selection wages and monopsonistic wages.

Anonymous - Telling monopsonistic wages and selection wages apart--On Claus Schnabel's comment
August 10, 2009 - 09:47

Thank you for you comment and your acute question. The problem of identifying one or the other mechanism has not occurred to me. I thought about it. A tentative argument is the following.

The monopsonistic theory ties wages to supply elasticities while assuming that labor is homogeneous in the ...[more]

... sense that all workers are equally productive. The selection wage theory ties wages to supply elasticities, too, but assumes that workers differ in productivity, and that this matters to the firms. This difference may help to identify which theory applies.

Labor heterogeneity would not matter for firms if productivity rests in the job. Think of a conveyor belt. All workers that meet minimum requirements are, from the point of view of the firm, perfect substitutes. Even if some workers could work much faster, this does not matter, as the speed of the production line determines the speed of production. For other jobs--such as salespeople or engineers--individual productivity differences matter for the firm. Such jobs exhibit wide "skill latitude." For jobs with pronounced skill latitude, heterogeneity becomes important.

Regarding identification of theories, an increase in mobility of women (more and better child care) would increase women's wages relative to male wages under both theories. An increase in skill latitude will increase selection wages but will leave monopsonistic wages unaffected. Hence we would expect under the scenario outlined in the paper, viz. lower supply elasticities of women, that sex discrimination would increase with increasing skill latitude, as it becomes more important to hire good workers, and this is easier to achieve for males.

All this is entirely hypothetical, as wage discrimination is illegal. Yet the same mechanism would work for regional discrimination. If technical progress aggravates the importance of skill differentials among workers while leaving supply conditions unchanged, we would expect an increase of wage differentials between urban and rural areas.

There is a further difference between the theories: With monopsonistic wages, firms are always on the supply curve, and all labor markets are cleared all the time. An increase in labor supply, brought about by reducing unemployment benefits, would reduce wages while maintaining cleared markets. On the other hand, selection wages will not necessarily be market clearing, and an increase in unqualified supply, brought about by a reduction of unemployment benefits, will not affect wages (see my Metroeconomica paper).

Anonymous - Referee Report
August 13, 2009 - 16:51

See attached file

Ekkehart Schlicht - Thank you
August 16, 2009 - 11:15

Thank you for taking the time and trouble to read the piece; thank you also for your positive evaluation and kind comments.

In case the paper gets accepted, I shall supply a revised version where I shall take up the points you mention: How to distinguish between monopsonistic and ...[more]

... selection wage mechanisms. I shall also point out that the selection wage story allows for uncleared markets, whereas the monopsonistic story does not.

Anonymous - Reader Comment
September 01, 2009 - 10:18

See attached file

Ekkehart Schlicht - Thank you
September 07, 2009 - 10:34

Thank you for taking the time for reading my submission, your positive evaluation, and you helpful suggestions which I shall take up.

a.) I shall follow your suggestion and add some remarks on related efficiency wage arguments (turnover, discipline, gift exchange).

b.) I ...[more]

... shall try to improve the section on social roles and shall check the definitions.

c.) I shall expand the notes on policy measures, and how they might affect discrimination.

Eskil Wadensjö - Reader Comment
September 03, 2009 - 15:03

See attached file

Ekkehart Schlicht - Florence's contribution
September 14, 2009 - 10:36

Thank you for commenting so extensively and sympathetically on my piece. I looked again at your book (which I read a long time ago, but unfortunately not when preparing the paper) and found some comments quite pertinent. I shall refer to them in a future version.

I ...[more]

... am particularly thankful that you drew my attention, through your book, to Florence's study (Economic Journal 1931). It predates Joan Robinson's contribution (which is usually cited) and contains a careful empirical discussion about the issue of differences in supply behavior between males and females. It seems largely forgotten and deserves to be referred to.

Ekkehart Schlicht - Counteracting Discrimination
September 14, 2009 - 10:47

I forgot to mention in the previous note that I find your comments on policy measures quite convincing and will take them up in a possible future version.

Anonymous - Referee Report 2
September 16, 2009 - 11:08

See attached file

Ekkehart Schlicht - Reply to second referee
October 09, 2009 - 11:55

See attached file.

Joyce Jacobsen - Decision Letter
September 18, 2009 - 09:01

This paper is very clearly written and thus easy to follow the argument therein. It has stimulated a good set of comments and dialogue with the readers, and thus has already served an important purpose. The two assigned referees and the other online readers have provided a large number of ...[more]

... useful comments for the author. I would only add two points: the third reader comment has a list of works which underscore that the monopsony idea is not recent; my recollection was that the Fanning Madden work was the first to apply the monopsony model to the gender case. This history of thought on the matter should be explicitly referenced in the paper. Second, one of the reasons why the Fanning Madden model didn't meet with wider acceptance was that many people argued that female labor supply, at least on the economy-wide level, was more rather than less elastic than male labor supply, and thus the model didn't seem applicable, or only applicable if one were also willing to argue that regional labor markets had more inelastic female labor supply (made plausible by tied mover arguments about family migration). The empirical evidence on female labor supply elasticity is dealt with only in passing in the article, and another reader comment explicitly calls the author to task for this matter. Overall though, my assessment is positive regarding this article and its expansion of the monopsony insight to the case where there are different distributions for both genders. Recommendation: accept, but encourage minor revision.

Ekkehart Schlicht - Reply to the Associate Editor's Decision
October 09, 2009 - 12:27

Thank you for dealing with the paper. I am glad, of course, that you found it acceptable.

As I understand your statement, you leave me quite some leeway to deal with the comments of the the referees and the readers. This is most welcome, as it would be ...[more]

... difficult to incorporate all suggestions in a comprehensive way. Useful as they are, such elaboration may hurt readability. So I shall try to find a good balance when rethinking yours and the other comments. As the author is interested in producing a paper that is as good as possible, I appreciate having obtained many useful suggestions without being advised about the way to deal with them in any specific way.

I have dealt the the points raised by the readers and the referees in my replies. Regarding your comments, I shall follow your suggestion and cite Fanning Madden. This is certainly appropriate. Further, I have done some searches and shall give further references to works that are largely forgotten to-day. Regarding the issue of supply elasticities on the firm level and the market level, I have in the meanwhile developed a fairly transparent way of formally demonstrating that supply elasticities, as seen by the firms, may differ from supply elasticities on the market level, such that market supply of women may be more elastic than that for males, while being less elastic from the firm's perspective. As this is a somewhat separate argument, I suggest to supply that in an appendix.

It may take some time to produce a revised version. Once I have completed it, I hope that you will perceive it as an improvement.

Ekkehart Schlicht - Revision
February 10, 2010 - 12:20

The revised version is attached.

I have benefited much from the comments of the associate editor, the referees, and the readers as well as from some other comments I have obtained. I would like to thank you all!

The main revisions are

1.) The argument ...[more]

... is related more more extensively to the literature on discrimination and also to the literature on the firm-size wage effect and on the determinants of inter-industry wage differentials.

2.) I have done extensive reading of the history of thought on the matter, unearthed interesting things that were new to me, but ultimately added only a few lines and some further references.
The additional references provided by the referees are very useful and have been incorporated.

3.) I have added a section on occupational and regional wage differentials in order to emphasize the point that the selection wage mechanism may be applied not only to sexual discrimination but to other wage differentials as well. This may help to put the overall approach as well as its particular application regarding discrimination in perspective.

4.) As a further illustration I have added an example illustrating that social stereotypes may affect wage differentials also in dimensions other than gender.

5.) Further, I have expanded and (hopefully) clarified the discussion of the informal discussion of the social multiplier and the channels.

6.) An appendix has been added that explains why market supply elasticities may differ from the supply elasticities faced by individual firms.

I apologize for the delay in preparing this revision. This is in part due to the fact that some of the earlier literature I wanted to screen was only available through inter-library loan. Further I waited for some comments (which proved very helpful in the end) quite a while.