Implications of relaxing the assumption of asymmetric information in Franco and Mitchell (2005)
Generally, an assumption of informational asymmetry often has two factors that seem to matter: the asymmetry itself, and the timing of the availability of information. In detail:
a) The case of no asymmetry- namely that it is common knowledge whether employee in question has learned something in his first period of employment or not – lies outside Franco and Mitchell’s exploration. It is clear that if their asymmetry assumption is being relaxed, then the implications of their theory change. One study has examined the nature of public information and optimal contracting by research institutions (Pakes and Nizan). In their inquiry they model publicly available information with ex post realization of learning outcome. The model implied that there are implementable first-best contracts (unlike the asymmetric information of Franco and Mitchell) that do not distort contracting equilibrium and profits. In that case, the CNCs (covenants not to compete) do not matter since they do not provide an additional enforcement mechanism necessary to move closer to the first best contract. However, Pakes and Nitzan use a mixed reward scheme composed of both monetary compensation and stock. It is seems that the stock compensation is necessary to produce a reward for the employee, which is high enough in order to prevent the employee from going to a competitor or forming a competing firm himself. This is mainly driven by uncertainty (according to Pakes and Nitzan) of invention’s future realization.
b) Relating to the timing of the available information, we will focus on two cases: 1) The ex ante information case and 2) the ex post information case.
1) In the case of ex ante information available to everyone, everything is simplified. One of the mathematical simplifications of the Franco and Mitchell model is that the incumbent firm’s profit is zero if the employee does not learn. Hence, hiring the employee which will not increase the profits will not be beneficial for any company and the optimal contract with this individual is no contract.
2) In the case of ex post information, both employer and the employee know that the employee learned in the first period. Thus, there would be some re-negotiation such that the employer offers a wage just high enough to prevent the employee from either going to the competitor or forming his own firm. In the instance that the re-negotiation is not possible the employer will offer a wage high enough to cover his (employer’s) expectation. Here, the stock scheme may be necessary (as in Pakes and Nitzan) to get to the contract that keeps the employee from leaving.
There are three possible extensions of this framework:
i) One may introduce uncertainty about the future realization of employee learning. This may cause a high enough distortion, such that the firm will not be able to offer the employee the optimal contract. In that case, CNCs may be necessary to force the employee to stay and we may very well see the results of Franco and Mitchell materialize again.
ii) The second extension is one in which the employer is not able to compensate the employee, not due to the uncertainty about future realization, but rather due to belief-differentials. Under that construction, the disagreements are likely to be a driving force behind employee’s choice to leave and future realization of these beliefs will provide different efficiency implications.
iii) The third possible direction is to introduce degrees of learning or differing degrees of employee ability. Here, one might think of a standard signaling story in which – ex ante – only the employee’s ability is known with some precision. Hence, the learning will be a function of ability and will not be known to the employer (he would only be able to estimate it based on the signal). In that case, high enough variance in the signals may also drive back the results to resemble those of Franco and Mitchell. Here, yet again, there may be a need for CNCs to move closer to the unachievable, otherwise first-best contract.
Instances of Production Knowledge Information Symmetry
The hypothesis proposed by Franco and Mitchell implies that if information asymmetry (regarding the knowledge that an employee has of the production technology developed by the firm) were reduced in an industry, that industry should show less agglomeration. This result is based on the crucial assumption made by the authors that the employer does not know whether his employee has learned the technology developed at the firm. Thus, if the assumption of asymmetry is relaxed
and CNCs are not allowed, the optimal contract becomes
. If the combined employer’s and employee’s profits given by the employee having learned and left the firm is less than the employer’s profit given by the employee having learned and stayed with the firm 
, and if the employer’s profit given by the employee having learned and staying with the firm is greater than the profits given by the opposite case 
, the employer has the incentive to offer an incentive for the employee not to leave. As long as 
holds, the employer will be able to offer compensation that eliminates the employee’s incentive to leave the firm. Thus, greater symmetry should reduce the number of spinouts, and hence, reduce industry agglomeration.
However, observation of agglomeration effects in industries with arguably greater symmetry than the Silicon Valley industry does not always imply this. The shoe manufacturing industry is one that does not require highly skilled labor. One can deduce that workers higher in the management hierarchy or with better training can easily learn any innovation within the firm, and the employer will have knowledge of this. Thus, the greater symmetry of information in the shoe industry should result in less agglomeration, and one that is diminishing over time. However, Olav Sorenson and Pino G. Audia show in their paper entitled “The Social Structure of Entrepreneurial Activity: Geographic Concentration of Footwear Production in the United States” that the agglomeration in the shoe manufacturing industry is significant and persistent through time.
A similar case is that of the watch manufacturing industry. Historically, the best watches have been those made by hand. This process involved highly trained laborers who were involved in most of the manufacturing process. Again, it is reasonable to conclude that an experienced watchmaker in a firm will learn any innovations, and the employer will know it. In this case greater symmetry should result in less agglomeration; however, in Europe, for example, the watch making industry has historically been concentrated in Switzerland.
One last case where we observe greater symmetry of information is in the automobile industry, particularly during its early period. During its initial stages, employees could easily learn innovations within the firm. One clear example of this was the set of knowledge spillovers relating to the assembly line. Henry Ford is usually credited with the invention of the assembly line; however, it was actually Ransom E. Olds, the founder of Olds Motor Vehicles Company, who invented it. Ford, who was actually a partner of Olds at some point and not an employee, copied the method and perfected it. Thus, we can conclude that at least in the early automobile industry, there was greater symmetry of information between employer and employees. However, as it is well known, the U.S. automobile heavily agglomerated in the Detroit area, and this agglomeration grew and persisted through time. Again, this fact contradicts the result that greater symmetry should result in less agglomeration.
Thus, we have that after relaxing the assumption of asymmetry made by Franco and Mitchell, less agglomeration should be the result in any industry, all other things being equal. However, three cases of industries with greater symmetry and highly agglomerated seem to contradict this result.
Data Sources for Labor Mobility
Future investigation to test for the different attributes of agglomeration will require a rich data set, particularly with regards to labor mobility. Rebitzer (2006) uses the U.S. Census Bureau Current Population Survey, but it is merely a monthly survey and lacks the advantages of a longitudinal data set. Another data set, The Longitudinal Employer – Household Dynamics (LEHD) Program at the Census Bureau, has a collection of infrastructure files that provide detailed information of workers, employers, and their interaction in the US economy. Since 2003, the Census Bureau has published the Quarterly Workforce Indicators (QWI). This is a new collection of data series that offers details on the local dynamics of labor markets across industries (http://lehd.did.census.gov/led/datatools/qwiapp.html).
For each state, there are data on total employment, net job flows, job creation, new hires, separations and turnover sorted by industry, year, sex, and age group. For example, in California, the net job flows in software publishers industry was -1,090 in the fourth quarter of 2004. If more detailed information is needed, the variables can be narrowed down over particular subsets of the data. For example, in the California example, we can narrow the set down to only male workers aged from 25 to 34. The net job flow was then -269. Quite significantly for the purposes of studying agglomeration, the data have information for 20 industries under which there are a number of selectable sub-industries. The LEHD serves as an excellent data source from which to construct general panel data on labor mobility.
References:
Burdett, Ken, and Melvyn Coles, (Sep 2003): “Equilibrium Wage-Tenure Contracts,” Econometrica,Vol. 71, No. 5. pp. 1377-1404
Franco, April F., and Matthew F. Mitchell “Covenants not to compete, labor mobility, and industry dynamics,” working paper, University of Iowa.
Rebitzer, James (2006): “Job hopping in Silicon Valley: The microfoundations of a high tech industrial district.” Review of Economics and Statistics, Vol. 88, No. 3, Pages 472-481.
Richard B. Freeman (1981) “The Effect of Unionism on Fringe Benefits”
Industrial and Labor Relations Review, Vol. 34, No. 4, pp. 489-509
James E. Long, Albert N. Link (1983) “The Impact of Market Structure on Wages, Fringe Benefits, and Turnover”, Industrial and Labor Relations Review, Vol. 36, No. 2, pp. 239-250
Olivia S. Mitchell (1983) “Fringe Benefits and the Cost of Changing Jobs”,
Industrial and Labor Relations Review, Vol. 37, No. 1, pp. 70-78
Dickens, F, Katz, A (1987) “Inter-industry Wage Differences and Industry Characteristics”, NBER WP: 2014
Pakes, Ariel and Saul Nitzan (1984), “Optimal Contracts for Research Personnel, Research Employment and Establishment of ‘Rival’ Enterprises,” Journal of Labor Economics, 1 (4), 345-65.