A classroom unemployment compensation experiment.
This experiment can be used in an introductory undergraduate macroeconomics course as a hands-on demonstration of the effects of unemployment compensation. During the course of the experiment, students observe how progressively higher levels of unemployment compensation affect unemployment rates, wages, and the distribution of income between skilled and unskilled workers. Because of its interactive nature, the experiment works best in classes of 16 to 50 students. However, the author and other professors have successfully run it in classes as small as 12 and as large as 90. It requires 45 minutes to run.
The experiment uses two labor markets: one for skilled labor and one for unskilled labor. The markets operate simultaneously. Students take the roles of workers and employers who negotiate employment contracts in these markets using a double oral auction. See Hazlett (1999) for a general description of how to administer a double oral auction market. The experiment consists of several periods, each of which represents hiring workers for one hour of labor. An employer can hire at most one worker from each market each period. Every period, workers try to earn the highest possible net benefits from working. Likewise, employers try to earn the highest possible net benefits from hiring them.
Workers calculate their net benefits of working by taking the difference between their wage and their cost of working. Different workers have different costs of working. Their costs reflect the various work-related expenses they face, such as transportation and day care, as well as the opportunity cost of their time. Similarly, employers calculate their net benefits by taking the difference between the revenue they earn from hiring someone and the wage they pay that person. Different employers earn different amounts of revenue.
Ready-to-use printouts of all the materials for running the experiment are available online by going to http://people.whitman.edu/~hazlett/econ/private, entering "private" as the username and "Modeling!" as the password. You will be prompted twice to enter the password. The materials available online include the instructions, the slips instructors pass out to assign students their roles in the experiment, the record-keeping sheets students use to track their actions, and follow-up discussion questions.
The experiment consists of two phases. In phase I, the government does not intervene in the labor markets. Phase I lasts about three periods, providing just enough repetition for the two markets to settle at or near their respective equilibria. In phase II, the government announces that it will offer unemployment compensation to any worker who chooses not to work. Some of the unskilled workers would earn smaller net benefits from working than the government offers in unemployment compensation. So they accept the government's offer and stop working. In subsequent periods of phase II, the government offers progressively higher amounts of unemployment compensation, causing more and more of the unskilled workers to join the unemployed. Only the unskilled workers with the lowest costs continue to work. In contrast, none of the skilled workers accept unemployment compensation. High demand for their labor services relative to their cost of working keeps their wages too high for them to be tempted by the government's offer.
As the rising unemployment compensation induces more and more of the unskilled workers to join the unemployed, the wages for the remaining unskilled workers increase. That is, the supply curve for unskilled labor shifts up and left as it incorporates the opportunity cost of forgoing the government's offer. This decrease in supply causes the wage to rise along the demand curve. The more generous unemployment benefits thus raise the earnings of the unskilled workers, narrowing the distribution of income between skilled and unskilled workers.
This experiment introduces a discussion of how the relatively low level of unemployment benefits in the United States might in part cause the relatively low unemployment rate and high level of income disparity we see in the United States. The U.S. economy resembles one of the early periods in the experiment, when the government offers low levels of unemployment compensation. Similarly, a western European economy such as Germany's, with its higher unemployment rate and tighter income distribution, resembles one of the later periods of the experiment, when the government offers more generous unemployment compensation. Thus, this experiment can serve as the opening for a discussion of the puzzling contrast between U.S. and western European unemployment rates and income distributions. Of course, there are many possible explanations for this contrast other than the differences in unemployment compensation programs highlighted in this experiment. For instance, alternative explanations include unionization or welfare-state institutions, such as employment protection.
2. Details on the Running of the Experiment
Before class begins, the instructor clears two separate areas of the room to serve as the trading floors for the skilled and unskilled labor markets. Employers will move between the two markets, while workers stay in their designated market. As students arrive, the instructor hands each a private information slip that establishes that student's role in the experiment. These slips assign workers their cost of working and employers their revenue from hiring. See Table 1 for examples of these slips. The instructor also passes out record-keeping sheets that the students will use to track their actions and net benefits.
The instructor then passes out the instructions and reads them aloud, inviting student questions. Once the instructor has announced the beginning of the first period, the employers and workers mingle, calling out wage offers. An employer will call out an offer to hire at a specified wage. Similarly, a worker will call out an offer to work at a specified wage. Either party may accept an offer from the other. After an employer and a worker agree on a wage contract, the worker comes to the front of the room to report the wage and identifications (ID) of the worker and employer. The instructor records this information on the blackboard (or uses a projector) for everyone to see. (Overhead slides or computer projection have the advantage of providing a permanent record of the results for later use.) The finished worker then sits on the edge of his or her trading floor to wait for the next period. The employer may go to the other labor market and offer to hire a worker there.
When trading activity ceases, the instructor asks whether the recorded results include everyone's trades. The instructor then requests and records the ID of anyone who did not trade. Table 2 provides an example of how the recorded results look. These particular results come from a session run in a Whitman College Principles of Economics class with 19 students. In this session, the theoretical equilibrium wage in the unskilled market ranges between $5.50 and $6.00. Unskilled worker A's cost of working lies above this equilibrium wage range, and employer 2's revenue from hiring an unskilled worker lies below this range. Worker A's unemployment motivates the government to announce its new unemployment compensation scheme.
Before period 4 begins, the instructor announces that the government will offer one dollar of unemployment compensation to any worker who does not work. When introducing the unemployment compensation scheme, the instructor explains that workers who do not work incur no cost of working. For instance, they need not pay transportation or day care costs.
The instructor also explains that workers who collect unemployment compensation will write on their record-keeping sheet the amount of unemployment compensation they receive that period. They will record this amount to the right of the column where they would have recorded their net benefit from working.
By expressing sympathy for the plight of unemployed workers such as worker A, the instructor provides a rationale for the government to begin offering unemployment compensation. However, students might question whether someone like worker A really qualifies as unemployed, given that he or she rejects all equilibrium wage offers. In fact, A is willing to work (at a wage of $6.50 or more), available for work, and actively seeking work. So A's situation does meet the official definition of unemployment. Later in the experiment, when other workers start rejecting wage offers in favor of collecting unemployment compensation, they too fit the definition of unemployed. They simply require higher wages than employers can afford to offer.
In period 5, the government increases its unemployment compensation offer by another dollar. Table 3 shows how the instructor records the results for this period. The results in Table 3, and all the results used as examples in this article, come from the same session with 19 students.
I recommend running about six periods in total. In phase I, the markets generally require two or three periods of repetition to reach equilibria or to get reasonably close. In phase II, the instructor runs periods with unemployment compensation levels of $1, $2, and $3. Each period requires about three minutes in a class of 20 students but about eight minutes in a class of 60. Most of the extra time in the larger classes comes from recording the results. An assistant's help recording the results reduces the length of a period considerably. Even so, the largest classes might not have enough time to run the period with $3 of unemployment compensation. Fortunately, the results with $2 of unemployment compensation are striking enough that the experiment could end after students see them. On the other hand, an instructor with extra time could run a period with unemployment compensation of $4, run other variations suggested at the end of this article, or start an in-class discussion.
Employers who cannot afford to hire a worker often ask if they too can collect unemployment compensation. In response, the instructor would explain that employers are searching for a marginal worker to add to their firm's existing personnel. The failure to find a marginal worker would not cause them to shut down their firm and become eligible for unemployment compensation.
For double oral auction experiments like this one, some instructors motivate students by rewarding their individual performances with extra-credit points or with prizes such as money or candy. Other instructors have a component of the course grade for good-faith participation in the experiment to encourage students to show up on time and stay engaged. These instructors find that their students show enough excitement at the opportunity to be active, plus a competitive desire to do well, that leaving the experimental profits hypothetical works fine. For this experiment, I use the latter method.
3. Parameters and Equilibria
When assigning roles to students, the instructor creates the same number of unskilled workers as employers. Having these numbers equal makes the unskilled wage and quantity predictable. It works best to have about half as many skilled workers as employers in order to make the skilled wage predictable. However, the number of skilled workers can vary considerably as long as it stays below the number of employers. The latitude in the number of skilled workers comes in handy when preparing for an uncertain number of participants. For instance, an instructor who has 34 students show up for the experiment can assign 14 to be employers, 14 unskilled workers, and 6 skilled workers. The instructor can then assign late-coming students the roles of skilled workers.
For the unskilled labor market, the experimental parameters have the following characteristics. As long as the instructor uses an equal number of employers and unskilled workers and has at least six of each, the unskilled labor market (absent government intervention) has a competitive equilibrium wage between $5.50 and $6.00. Plus, there will always be at least one unskilled worker unemployed, even with no unemployment compensation program in place. Large classes have more of these initially unemployed unskilled workers in proportion to the size of the class. The private information slips available online work for classes as large as 100.
It is possible to run the experiment with a class as small as 11 students by having five employers, five unskilled workers, and one skilled worker. In this case, the unskilled wage ranges between $5.50 and $6.50 in equilibrium, absent government intervention.
For the skilled labor market with N workers and more than N employers, the experimental parameters have the following characteristics. All N skilled workers will be hired at a competitive equilibrium wage (absent government intervention) that ranges between the revenue for the Nth and (N + 1)th highest-revenue employer. For instance, consider our example class of 19 with its five skilled workers. Tables 4 and 5 show the workers' costs and employers' revenues for this class. In equilibrium, the skilled wage equilibrium ranges between $18.00 and $18.50 because the fifth highest-revenue employer (employer 1) could afford to pay $18.50, whereas the sixth highest-revenue employer (employer 5) could afford only $18.00. So employer 1 outbids employer 5 to get the last available worker.
Figure 1 shows the supply and demand curves in the unskilled labor market for the class of 19. Figure 2 does the same for the skilled labor market. Each of these figures shows both the supply of labor with no unemployment compensation and the supply of labor with $2 of unemployment compensation. In each market, the labor supply curve has shifted up by $2 because forgoing the government's unemployment compensation offer raises the opportunity cost of working by $2. Note that in the unskilled labor market, unemployment compensation raises the equilibrium wage and reduces the quantity of labor hired. However, in the skilled labor market, unemployment compensation changes neither the equilibrium wage nor the quantity of labor hired. The intense competition for the relatively small amount of skilled labor keeps their equilibrium wage within the $18.00 to $18.50 range and would do so even if unemployment compensation rose to $4. So with the parameters in this experiment, rising unemployment compensation reduces the earnings gap between the skilled and unskilled.
[FIGURES 1-2 OMITTED]
Table 6 shows the theoretical outcomes for the unskilled labor market in the class of 19. Table 7 shows relevant actual results from the skilled and unskilled markets. Note that none of the skilled workers ever becomes unemployed.
There are some anomalies in these results. In period 2, worker E is unemployed even when working would presumably provide greater net benefits. In periods 4 and 5, worker D accepts a job even though unemployment compensation exceeds the net benefit from working. It is possible that both students suffered from some early-on confusion that resolved itself later. It is also possible that philosophically D would rather work than go on the dole, at least until the dole becomes fairly large. Certainly, confusion or a philosophical preference for work could both occur in the real world as well.
Despite the anomalies, the experimental results do show that higher levels of unemployment compensation induce more of the unskilled to join the ranks of the unemployed. We also see wages for the unskilled rising as the level of unemployment compensation increases, narrowing the gap between the earnings of the skilled and unskilled.
The following questions work in a follow-up class discussion. Descriptions of typical student answers appear after each question. The following section includes information that an instructor could briefly describe to students to encourage them to think about and comment on possible connections between the experimental results and labor market policies.
(i) What did you (the workers and employers) think of unemployment compensation?
Workers often express their approval of the unemployment compensation scheme or at least what it did for their bottom line. Workers do sometimes express disapproval of what unemployment compensation did to their work ethic. Employers uniformly disapprove of the unemployment compensation scheme. They report earning lower net benefits, both due to paying higher wages and to hiring fewer workers.
(ii) In our experiment, what effect did unemployment compensation have on unemployment rates and wages?
Students note that higher levels of unemployment compensation generated both higher unemployment rates and higher wages among the unskilled. However, unemployment compensation did not affect the employment of the skilled workers. Nor did unemployment compensation seem to have an effect on the wages of the skilled workers.
(iii) Why did some unskilled workers drop out of the job market to accept unemployment compensation?
The workers explain that employers were not willing to pay enough to keep them in the market, to which employers respond that they could not afford to.
(iv) What do you think the government could have done to help unemployed workers without further reducing the number of people employed?
Employers typically respond that a government subsidy for hiring workers would make hiring more of them affordable. Other student suggestions include lowering workers' costs by providing services such as public transportation. The instructor could follow up on these comments with an example of how work-related expenses can make accepting a job too expensive for a potential worker. For instance, in the case of a central-city resident who does not own a car, a service job in the suburbs might constitute this person's best job prospect. However, the typical hub system of public transportation is set up to take workers from homes in the suburbs to jobs in the central city, not the other way around.
Some Additional Information for Discussion
To promote further policy discussion, the instructor could ask students to imagine that each period of the experiment corresponds to a situation in a different country. For instance, a period with only $1 of unemployment compensation corresponds to a country like the United States, with its relatively low level of unemployment compensation. A period with a higher level of unemployment compensation corresponds to a country like Germany, with its more generous unemployment compensation program. In particular, in the United States, unemployment compensation normally lasts at most six months, replaces only 50% of lost wages, and is difficult to qualify for in that a person must work before being eligible to collect and cannot have voluntarily quit or been fired for cause. In contrast, in a country like Germany, benefits last longer (almost indefinitely with token worker retraining programs), replace a greater percentage of wages, and are easier to qualify for. In fact, people can collect without ever having held a job. See Solow (2000) for more information comparing U.S. and European labor markets.
In the experiment, periods with higher unemployment compensation levels had higher unemployment rates and narrower distributions of income. So again, these periods resemble the situation in a western European country such as Germany. In fact, Germany suffers from unemployment rates almost twice that of the United States, and it has a much narrower distribution of income. (1) De Nardi, Ren, and Wei (2000) report that the ratio of the average labor earnings of the richest and poorest 20% of households is 21 in the United States but only 12 in Germany.
The instructor then asks students for their impression of whether the experiment results suggest a connection between the level of unemployment compensation, the unemployment rate, and the degree of income disparity in a country. In particular, could the less generous unemployment compensation we have in the United States (like the period in the experiment with unemployment compensation of only $1) account for the lower unemployment rate and higher degree of income disparity in the United States than in some continental European countries?
Finally, the instructor can point out how technological change and the opening of international markets have lowered demand in the developed countries for unskilled versus skilled labor. Some economists, for instance, Mankiw (2003, pp. 172-3), suggest that over the past two decades the result in the United States of falling demand for unskilled labor differs widely from the result in Europe. In the United States, unskilled wages fell relative to skilled wages, producing greater income disparity. In European countries, where unemployment compensation provided a more generous safety net, higher unemployment resulted instead.
Data Analysis Questions
The following questions ask students to analyze in depth the data generated in the experiment. These questions can promote a more detailed discussion in a later class meeting or serve as a laboratory report assignment. To answer these questions, students need information on all the employers' and workers' reservation wages, as in Table 4. They also need the experimental results in the form of a list of the wage contracts, including ID, each period.
(i) Suppose that the government offers no unemployment compensation. Graph the supply curve for labor in the unskilled labor market. (Hint: You find this labor supply curve from the unskilled workers' costs of working, not from the results of the experiment.) On the same graph, show the demand curve for labor in the unskilled labor market. (Hint: You find this labor demand curve by using the employers' revenues from hiring unskilled workers, not from the results of the experiment.) What is the theoretical competitive equilibrium wage and quantity of labor hired in the unskilled labor market? What does theory predict will be the gains from trade for employers and workers in the unskilled labor market?
(ii) Suppose again that the government offers no unemployment compensation. On a separate graph from that of question 1, graph the supply and demand curves for labor in the skilled labor market. (The hints from question 1 apply here too.) What is the theoretical competitive equilibrium wage and quantity of labor hired in the skilled labor market? What does theory predict will be the gains from trade for employers and workers in the skilled labor market?
(iii) How do the experimental results in the early periods (before the government offers unemployment compensation) compare to the theoretical predictions you described in questions 1 and 2?
(iv) Consider the situation when the government offers unemployment compensation of $2. What happens to the supply curve for labor in the unskilled market? What happens to the supply curve for labor in the skilled market? Graph the new supply and demand for labor in the unskilled market. On a separate graph, show the new supply and demand for labor in the skilled market. What does economic theory predict happens to the wage and quantity of labor hired in each market? Does theory predict that the labor markets will efficiently allocate resources given the government intervention?
(v) How did the experimental results compare to the theoretical predictions you described in question 4?
(vi) Who benefited from having a high level of unemployment compensation? Was anyone hurt by having high unemployment compensation? How much did the government spend on unemployment compensation in each period?
(vii) How did the level of unemployment compensation affect the wages earned by unskilled workers versus the wages earned by skilled workers? That is, how did the level of unemployment compensation affect the distribution of income between the skilled and unskilled?
(viii) Consider the following observations. In some western European countries such as France and Germany, unemployment benefits are much more generous than in the United States. The United States has much lower unemployment rates than France or Germany. The United States also has a much wider distribution of income than either of these countries. That is, there is a bigger gap between the earnings of the skilled and the unskilled in the United States than in western Europe. Can the results from this experiment help explain these observations?
(ix) If the government funded unemployment compensation with a tax on workers' incomes, how would unemployment compensation affect the labor market? If this tax were paid by employers, what would happen to the labor market?
5. Variations Suggested by Students
In some sessions, the unskilled workers spontaneously requested a subsidy for working rather than compensation for not working. Similarly, I have had students request that we subsidize employers for each unskilled worker they hire. I have also had students request that we institute a minimum wage. In all these cases, students were concerned about what they could see unemployment compensation doing to the labor market. If time permits, I recommend letting the class experiment with suggestions like these. For instance, in one session, at the students' request, I instituted an impromptu $2 reduction in the cost of working for all workers. I described this cost reduction as the result of the government paying for public transportation, thereby reducing workers' transportation costs. Unemployment compensation did not exist while the cost reduction was in effect. (I let the unskilled workers choose the amount of the reduction. In fact, their choice of the $2 happened to just allow everyone to profitably accept work.) With this public transportation program in place, worker A could afford to accept a job for the first time and did so. However, he earned a net benefit of only $0.25. He therefore lobbied the government to end the public transportation program and return to the unemployment compensation program, which we did in the subsequent period. Unemployed again, he received $3 in unemployment compensation and was much happier.
There exists no pat answer to why U.S. and continental European unemployment rates and income distributions differ. Some of the many potential answers include differences in the power of labor unions or differences in labor market regulations, such as restrictions on discharging workers. This experiment promotes discussion of another possible cause: differences in unemployment compensation programs. The experiment generates data that students can analyze to test the hypothesis that more generous unemployment compensation causes a higher unemployment rate and a narrower distribution of income.
Table 1. Sample Private Information Slips for Workers and Employers in Each Market You are worker A, in the unskilled labor market. Your cost of working for an hour is $6.50. You are employer 1. The revenue you earn from hiring a skilled worker to work for an hour is $18.50. The revenue you earn from hiring an unskilled worker to work for an hour is $7.00. You are worker V, in the skilled labor market. Your cost of working for an hour is $12.50. Table 2. Example of the Results Recorded on the Blackboard or Overhead by the Instructor Period 3 Skilled Market Unskilled Market 6 hires V at $16.75 5 hires B at $5.00 3X $18.00 4G $6.00 1W $18.50 6E $5.50 4Z $17.00 1F $6.00 7Y $17.50 3D $6.00 7C $6.00 Did not hire: 2, 5. Did not work: none Did not hire: 2. Did not work: A Table 3. Example Results Recorded on the Blackboard for a Period in Which the Government Offers Unemployment Compensation Period 5: The government will give any worker who chooses not to work $2 in unemployment compensation Skilled Market Unskilled Market 4Z $18.00 7E $6.50 3Y $18.00 1B $6.25 1W $18.25 4G $6.00 7V $17.75 5F $6.50 6X $18.00 3D $6.25 Did not hire: 2, 5 Did not hire: 2, 6 Unemployment compensation recipients: A, C Table 4. Workers' Costs for a Class of 19 Unskilled Cost of Skilled Cost of Worker ID Working ($) Worker ID Working ($) A 6.50 V 12.50 B 3.50 W 11.50 C 4.50 X 12.00 D 5.50 Y 11.50 E 4.00 Z 11.50 F 5.00 G 2.50 Table 5. Employers' Revenue for a Class of 19 Revenue from Employer ID Hiring a Skilled Revenue from Hiring Worker ($) an Unskilled Worker ($) 1 18.50 7.00 2 20.00 5.00 3 19.00 6.50 4 19.50 7.50 5 18.00 8.00 6 18.50 6.00 7 18.00 8.50 Table 6. Theoretical Equilibrium Outcome in the Unskilled Labor Market Unemployment ID of the Equilibrium Gains from Compensation Unemployed Wage ($) Trade ($) $0 A 5.50-6.00 18.50 $1 A, D 6.00-6.50 18.00 $2 A, D, F 6.50-7.00 16.50 $3 A, C, D, F 7.00-7.50 14.00 $4 A, C, D, E, F 7.50-8.00 10.50 Table 7. Actual Results from the Skilled and Unskilled Labor Markets Unemployment Average Average Compensation ID of Unskilled Skilled Period Policy Unemployed Wage ($) Wage ($) 1 None A 4.90 15.40 2 None A, E 5.35 16.65 3 None A 5.75 17.45 4 $1 A 5.90 18.00 5 $2 A, C 6.30 18.00 6 $3 A, C, D, F 7.10 18.05 7 $4 A, B, C, D, E, F 8.00 18.25
Support for this work was provided by the National Science Foundation's course Curriculum and Laboratory Improvement Program under grant DUE-9950688. The author is grateful to Jim Shepherd and Halefom Belay of Whitman College, Pare Whalley and Ta Herrera of Western Washington University, Steve Perez of Washington State University, and Noelwah Netusil of Reed College for letting her run versions of this experiment in their classes and to Michael Haupert of the University of Wisconsin. La Crosse; John Duffy of the University of Pittsburgh; Earl Adams of Allegheny College; and Linda Hooks of Washington and Lee University for beta-testing the experiment. Likewise, she appreciates the helpful suggestions of participants in seminars at Washington State University and the University of Otago.
(1) For international data on unemployment rates, see Federal Reserve Bank of St. Louis (2003).
De Nardi, Mariacristina, Liqian Ren, and Chao Wei. 2000. Income inequality and redistribution in five countries. Federal Reserve Bank of Chicago Economic Perspectives 24(2):2-20.
Federal Reserve Bank of St. Louis. 2003. International economic trends. Accessed 29 May 2003. Available http://research. stlouisfed.org/publications/iet.
Hazlett, Denise. 1999. Economic experiments in the classroom. New York: Addison-Wesley.
Mankiw, N. Gregory. 2003. Macroeconomics. 5th edition. New York: Worth Publishers.
Solow, Robert. 2000. The European unemployment problem. CESifo Forum 1(1):3-5. Munich: Ifo Institute for Economics Research.
Denise Hazlett, Department of Economics, Whitman College, 345 Boyer Avenue, Walla Walla, WA 99362, USA; E-mail firstname.lastname@example.org.
|Printer friendly Cite/link Email Feedback|
|Title Annotation:||Targeting Teaching|
|Publication:||Southern Economic Journal|
|Date:||Jan 1, 2004|
|Previous Article:||Comparing student achievement across experimental and lecture-oriented sections of a principles of microeconomics course.|
|Next Article:||The Economics of the World Trading System.|