The Basis of International Trade: A Classroom Experiment Alberto Isgut, Ganesan Ravishanker, and Tanya Rosenblat * Wesleyan University Abstrat We introdue a simple web-based lassroom experiment in whih students learn the Riardian model of international trade. Students are assigned to ountries and then make individual prodution, trade and onsumption deisions. The analysis of experimental data introdues students to the onepts of absolute and omparative advantage, relative pries, prodution possibility frontier, speialization, gains from trade, utility maximization and general equilibrium. Students learn about the relationship between individual deision-making and aggregate eonomi ativity. The assoiated software, Riardian Explorer, is easy to setup and requires minimal preparation time for instrutors. The game is developed as a tool to omplement ourses in international trade, but it an be used in introdutory and intermediate miroeonomis ourses as well. The analysis of teahing effetiveness has demonstrated that integration of this experiment in the urriulum enhanes student learning. Keywords: Absolute advantage, omparative advantage, speialization, prodution possibility frontier, gains from trade, utility maximization, general equilibrium, lassroom experiments First version: January 2005 This version: April 2005 We thank Wesleyan University for finanial support; Mike Lovell and poster session partiipants at the 2005 meetings of the Amerian Eonomi Assoiation for suggestions; Manolis Kaparakis, Mike Roy, Henk Meij, and Anne Loyer for their partiipation in different stages of this projet; John Hammond for tehnial assistane; and our students of the last three years for their enouragement and onstrutive ritiism. Isgut (orresponding author): Department of Eonomis, Wesleyan University, Middletown, CT 06459, aisgut@wesleyan.edu, Tel: (860) 685-3958, Fax: (860) 685-2301; Ravishanker: ITS, Wesleyan University, ravishan@wesleyan.edu; Rosenblat: Dept. of Eonomis, Wesleyan University, trosenblat@wesleyan.edu.
4/5/2005 The basis of international trade: A lassroom experiment page 1 I. Introdution David Riardo s priniple of omparative advantage, whih states that a ountry gains from trade by speializing in the prodution of goods that it is relatively better at produing, is familiar to eonomis students. The Riardian model of trade lays the foundation of the modern theory of international trade and is usually taught in introdutory eonomis ourses. Comparative advantage, however, is not an easy onept (Krugman, 1996). Paul Samuelson named it the most ounterintuitive yet ompelling proposition in the soial sienes (Bhagwati, 2003, p. 5), and Paul Krugman (1993) emphasized its relevane in the undergraduate eonomis urriulum. To failitate the intuition, the Riardian model is taught in introdutory ourses with the aid of graphs depiting the prodution possibilities frontier in the two ountries engaged in trade. In International Trade ourses, equations and further graphial analysis are added. For example, the Krugman and Obstfeld (2003) textbook introdues a relative demand-relative supply graph to illustrate the determination of equilibrium in the goods markets and a labor market graph to illustrate equilibrium in the Riardian model with more than two goods. While the graphial analysis is a very useful pedagogial tool, some deeper elements of the model, suh as the onept of general equilibrium and the inter-relation between miro deisions and aggregate market outomes, are often overlooked in standard textbook treatments. We believe that a pedagogial tool in whih students experiene the deision problem of eonomi agents and observe the results of their interations with other agents an help them appreiate more fully these deeper elements. In this paper we desribe how to use web-based lassroom experiments to failitate and enhane learning of the Riardian model. Students are assigned to ountries and are endowed with labor whih they an use for the prodution of goods. Students make individual deisions on what goods to produe, whether to engage in trade and what goods to offer for trade. Trade is
4/5/2005 The basis of international trade: A lassroom experiment page 2 implemented using a posted offer institution in whih sellers enter their offers and buyers have an opportunity to aept desirable trades. After prodution and trade, students make their onsumption deisions. Analysis of the resulting data makes students think arefully about the relationship between their individual deisions and resulting aggregate eonomi outomes. They grow to appreiate the role of pries in oordinating eonomi ativity. The game is played using Riardian Explorer, a software pakage designed at Wesleyan University and available for use at other institutions free of harge 1. It allows for a larger number of partiipants in a given experiment than traditional paper and penil experiments. 2 Therefore it is attrative both for instrutors who already use paper and penil experiments and for those who are onsidering adoption of lassroom experiments in their urriulum. 3 The program was originally designed for an International Trade ourse, but is appliable more broadly for teahing key eonomi onepts like opportunity osts, relative pries, prodution possibility frontier, absolute and omparative advantage, speialization, gains from trade, utility maximization, and general equilibrium at various levels of miroeonomis. Sine all transations are onduted as barter exhange, the program an be also used in a monetary eonomis ourse to study the origins of ommodity money. Finally, the program an be adopted in ourses in experimental eonomis to study posted offer markets. We have suessfully used the software at Wesleyan as part of an introdutory urriulum and also in field ourses in international trade. Finally, we enourage onduting the experiment prior to assigning textbook readings or doing formal analysis in lass. This allows students to learn the onepts experientially. Playing the game (with different parameters) for the seond time after lass disussion and related homework reinfores the aquisition of the key onepts and also inreases student satisfation beause they feel more in ommand of the material. The remainder of the paper proeeds as follows: In setion II, we summarize the underlying theoretial model. In setion III, we explain the proedures for onduting the
4/5/2005 The basis of international trade: A lassroom experiment page 3 experiment in a lassroom setting. Setion IV presents a disussion of the results of a version of this experiment implemented at Wesleyan in the Fall of 2004. In this ontext, we highlight some of the key pedagogial points that the experiment an be used to address. Setion V provides evidene of teahing effetiveness using a ontrolled evaluation, and setion VI onludes. II. Riardian model The following desription of the Riardian model emphasizes the inter-relation between miro deisions and aggregate market outomes. Eah ountry is populated by I prodution and onsumption units indexed by i. Think of them as family farms endowed with a fixed amount of labor L i to use during eah prodution period. All farms within a ountry have aess to the same prodution tehnologies, whih are linear in labor, the only fator of prodution. Assuming for simpliity that there are only two goods, x and y, ountry s prodution funtions are x = ( 1 a )L and y = ( 1 a )L, LX LY where a LX and a LY are the unit labor requirements parameters. The prodution possibilities frontier (PPF) desribes the maximum levels of prodution of x and y available to the farm when it utilizes its labor endowment fully: L = a x a y. (1) i LX + LY Eah farm an produe for its self onsumption or for the market. In this simple eonomy there is no monetary unit: all market transations are done through barter. Using y as numeraire, the market value of a farm s prodution basket ( x p, y p ) is defined as i p p P P ( x y ) p ( p p ) x y V, +. (2) The onsumption possibilities frontier (CPF), Y p p ( x y ) p ( p p )x X Y y = V, (3) i Y X Y
4/5/2005 The basis of international trade: A lassroom experiment page 4 desribes the maximum levels of onsumption of x and y available to the farm when it trades its prodution basket ( x p, y p ) in the market. In deiding whether or not to engage in a market transation, the farm must ompare its PPF with its CPF. In the example in Figure 1, the relative market prie of x exeeds the farm s opportunity ost of produing x, a LX a LY, making it profitable for the farm to sell x in exhange for y. Beause all farms in a ountry have the same opportunity osts, the autarky equilibrium prie is p p = a a. Therefore, in autarky farms have no inentive to produe for the X Y LX LY market. Under free trade, farms will have an inentive to trade with farms in other ountry provided that opportunity osts differ aross ountries. Suppose that there are only two ountries = 1, 2 and that ountry 1 has a omparative advantage in good x: a a < a a. Farms 1 LX 1 LY 2 LX 2 LY in ountry 1 will have an inentive to sell x in exhange for y if a a < 1 LX 1 LY p X p Y, and farms in ountry 2 will have an inentive to sell y in exhange for x if 1 1 2 2 the free trade equilibrium prie should lie in the [ a a a ] LX a a < p 2 LY LY, LX LY 2 LX Y p X. Therefore, a interval. Any transation at a prie outside this interval will be prejudiial to one of the parties involved. To reap, the model desribed so far has preditions at both the miroeonomi and the aggregate level. At the miroeonomi level it predits that farms will not sell at a relative prie below their opportunity ost, nor will they buy at a prie above their opportunity ost. It also predits that farms will have an inentive to speialize in the prodution of the good whose relative prie exeeds its opportunity ost, and to sell that good in the market in exhange for the other good. At the aggregate level it predits the diretion of trade flows and a range of pries within whih the equilibrium prie will lie. Given the linearity of the prodution tehnologies, the model also predits that at least one of the ountries will be fully speialized in the prodution of the good in whih it has a omparative advantage. Finally, when only one ountry fully
4/5/2005 The basis of international trade: A lassroom experiment page 5 speializes, the model predits that the equilibrium prie will equal the opportunity ost of the ountry that does not fully speialize. The remaining ambiguity about this model is what exatly will be the free trade equilibrium prie. If one wants that prie to be unique, the model an be augmented with assumptions about preferenes and onsumption hoies. The disussion so far has impliitly assumed that farms don t want to lose from trade. To be more preise we an assume that eah family farm hooses a olletive onsumption basket to distribute among its members by maximizing a utility funtion. The usual assumption is that the utility funtion is quasi-onave, homotheti, and idential aross all farms in all the ountries. An example of suh utility funtion, whih we use in the Riardian Explorer game is the CES utility funtion 1 ρ ρ ( x + y ) ρ, < 1 U ( x, y) = ρ. (4) This funtion inludes the logarithmi and the Leontief utility funtions as speial limiting ases, when ρ approahes 0 and, respetively. III. Proedures In this setion, we briefly desribe the proedures for setting up an experiment, review sample game sreens and outline the sequene of user ations. 1. Software Requirements The software is a Java appliation that was developed by Information Tehnology Servies (ITS) at Wesleyan and runs on a Wesleyan server. All users, inluding the instrutor only need to have internet aess through a web-browser (suh as Internet Explorer). The browser needs to be Java enabled running Java 2 v. 1.43.x. or later. Users who do not have Java enabled browsers an download Java plugin for free from http://www.java.om
4/5/2005 The basis of international trade: A lassroom experiment page 6 Instrutors an set up their own game or use pre-programmed games. There is no limit on the number of users; therefore the software an be used even in very large lasses. Users from outside of Wesleyan need to set up an aount for authentiation prior to using the software. The instrutor an do this by ontating ITS and supplying a spreadsheet with student names and emails. The ontat information is available at the game web site, http://www.wesleyan.edu/re. The instrutor an view the game using the experimenter ontrol sreen. After the game is over, the data an be downloaded in a omma-delimitted file and then opened using a spreadsheet program suh as Exel. The software also generates basi summary statistis and graphs that an be viewed on the sreen, printed out or projeted during leture using multimedia equipment. 2. Setting up a game As we mentioned, instrutors an hoose among the existing games or reate a new one. A game is defined by hoosing ountry names, good names, and parameter values for total available labor, unit labor requirements and ρ, the utility funtion parameter. 4 While the software has no limit on the total number of ountries and goods, we suggest using a simple example with two ountries and two goods in introdutory ourses. This provides a natural parallel to the textbook disussion and also does not overwhelm students with omplexity. If students have not been exposed to utility maximization, we suggest using Leontief preferenes beause they are more transparent to explain. 5 In addition, the game allows the option of arrying forward goods not onsumed in a round to the next round, subjet to a good-speifi storage ost set by the instrutor. Besides hoosing theoretial parameters, instrutors also set up when the game will be played, its duration, and the number of trading rounds. 6 During eah round students make prodution deisions, offer goods for trade, aept offers from other students, and deide how muh to onsume. The game an be set up suh that there is time between eah round for
4/5/2005 The basis of international trade: A lassroom experiment page 7 students to review the outomes of the previous rounds and design a strategy for the next round. When deiding on the length of the experiment, the instrutor has to deide on the length of eah round and the time between rounds. For a 50 minute lass, we suggest setting up a game with 4 rounds of 8 minutes in length with 4 minutes in between to review results. Instrutors an also set up a game to run outside of lass. There are two options for doing this. One is to require all students to play at a speified time from their own omputers. The other is to set up a game lasting for several days. We prefer the first option beause the markets an be very thin if students do not login at the same time. Students are assigned to a ountry as they login. In the urrent version, the number of students assigned to eah ountry is similar. However, it is possible to make ountries different in size by hoosing different values of the labor endowment parameter aross ountries. The instrutor does not need to worry about assigning students to a ountry as it is done automatially by the program. 3. Proedures for users Students are asked to login into the game using a Java-enabled browser. Upon login they are automatially assigned to a ountry and immediately see the game sreen (Figure 2). The game sreen used in the Riardian Explorer web site is programmed as a Flash appliation suh that students an see a desription of the funtion of eah partiular area of the sreen when they bring their mouse to that area. This provides an easy way for students to get quikly aquainted with the interfae. Instrutors an hoose to show the game sreen to students in advane and answer any questions they might have in order to minimize onfusion during the first round of play. The game information portion of the sreen tells a player whih ountry she was assigned to, her initial labor endowment and also traks her utility in the urrent round and throughout the game. The top left-hand orner of the sreen traks the time remaining in a round and available labor. Next to it is information on eah good inluding unit labor requirements for prodution,
4/5/2005 The basis of international trade: A lassroom experiment page 8 the number of goods produed in the urrent round, the number of goods bought/sold in the urrent round, the number of goods onsumed and the number of goods available for onsumption or trade. The middle portion of the sreen is reserved for student ations. On the left, they hoose how muh of eah good to produe, onsume and offer for trade. All trades are presented as barter exhanges (good 1 in terms of good 2). Offers urrently available on the market are displayed on the right side of the sreen. On the top student sees her own offers, whih she an hoose to retrat. Below are all offers on the market that she an hoose to aept. The offers are sorted first by the good offered and then by the relative prie. Eah student begins a round by making prodution and trade deisions. At the end of the round, students hoose to retrat outstanding offers and make their onsumption deisions. There is a shortut button that allows them to onsume everything they are holding. Their utility sore is omputed. After the round is over, students spend several minutes reviewing the results and hoosing a strategy for the next round. In eah round they get a new alloation of labor. There is an option (hosen by instrutor) to allow students to arry over goods from one period to the next. This an be espeially attrative if one hooses to use the program to study the origins of ommodity money. Students an learn how storage ost properties of different goods enourage their development as a medium of exhange. Notie that when this option is not hosen, all the goods in stok not onsumed before the end of the round is lost. Currently the sore is omputed with utility at autarky normalized to one, with any numbers above indiating gains from trade. We are also in the proess of developing an alternative game sore based solely on gains from trade. This will allow instrutors to hoose whether or not to inlude utility maximization as a teahing objetive for this experiment. The game sreen is automatially refreshed every ten seonds and reflets all prodution, trade and onsumption deisions. Students an rejoin the game at any time by opening the browser and typing in their login information. This is onvenient in ase they experiene omputer problems or lose their browsers by aident. Another attrative feature of the software
4/5/2005 The basis of international trade: A lassroom experiment page 9 is its messaging apability. Students an send short text messages to eah other, and the instrutor an send messages to individual players or to all partiipants. Messages an be used by the instrutor to make omments about the development of the game and by students to build trade relationships. Students find this feature very valuable. IV. Classroom Disussion and Analysis of Sample Data In this setion we show data for a sample game to illustrate the pedagogial possibilities of the Riardian Explorer game. This game, alled Riardos, was played by international trade students at Wesleyan University in the Fall of 2004. The game is haraterized by the following parameters: Unit labor requirements Labor Country Wine Cloth Endowment England 2 3 10,000 Portugal 1 6 10,000 Rho -2 Notie that Portugal has omparative advantage in wine and England in loth. Of ourse, players who log in to the game do not know this a priori; they only know the parameters orresponding to their respetive ountry. Ideally, as the game progresses and players observe pries of trades in the market, they realize in whih good they have a omparative advantage and start to take prodution deisions aordingly. Figure 3 shows prodution hoies at the end of the 2 nd and 4 th rounds of play for players loated in Portugal. The vertial axis represents wine and the horizontal axis represents loth. It is lear that between the 2 nd and 4 th rounds students learned to produe at the PPF. Moreover, in the 4 th round 6 six out of twelve students fully speialized in wine, as the theoretial model predits, ompared to three out of thirteen in the 2 nd round.
4/5/2005 The basis of international trade: A lassroom experiment page 10 Figure 4 shows onsumption hoies of the same Portugal players at the end of the 2 nd and 4 th rounds of play. It is lear that more players managed to gain from trade and onsume above their PPF, as the model predits. But it is also lear that many players onsumed even less than what they produed. Moreover, the graph does not show that four players in the 4 th round and five players in the 2 nd round did not onsume at all. When we asked students about zero onsumption hoies, some told us that they were so involved in the trading that forgot to lik the onsume button before the end of the round. Beause this partiular game does not allow arrying forward goods to the next round, they lost all the goods they produed or aquired through trade. Nevertheless, when looking at the data generated by the game, we ould verify from the students unonsumed inventory at the end of the 4 th round that three out of the four with zero onsumption ould in fat have onsumed outside the PPF if they had liked the onsume all button. This problem suggested to us the need to improve the feedbak to players during the game. We believe that the new graphial interfae under development depiting the PPF, the indifferene urves, and the onsumption and prodution points of the player at eah point in time will be helpful both as a pedagogial tool, and to improve students experiene playing the game. An important predition of the Riardian model, as we have mentioned in Setion II is that market pries should lie on the interval determined by the two ountries opportunity osts. This means that the prie of wine should be in the [1/6, 2/3] interval and the prie of loth should be in the [1.5, 6] interval. A more preise predition an be obtained if eah player hooses wine and loth to maximize his or her utility funtion. Given the parameters of this game, the theoretial equilibrium prie of wine in terms of loth is 1/6, whih equals Portugal s opportunity ost of wine in terms of loth. That is, in this game, only players loated in England should fully speialize and gain from trade. What does the data say about these preditions? Figure 5 below shows the pries of transations in the wine and loth markets, respetively. In the Riardian Explorer game a player
4/5/2005 The basis of international trade: A lassroom experiment page 11 an partiipate in a market by posting an offer or by aepting a posted offer. For example, in the wine market players post or aept offers of wine in exhange for loth. The data shows a reasonable degree of onvergene of pries to the theoretial range in the wine market. Moreover, towards the end of the game, there is a greater onentration of observations around the lower bound, 1/6, as predited by the model when it inludes utility maximizing behavior. The loth market, instead, shows more observations outside the expeted range. However, towards the end of the game there also seems to be a onentration of observations around the theoretial value, whih in this ase is the upper bound of 6. To further investigate players hoies in the game, Table 1 omputes transation errors in the two markets during the four rounds of play. One type of error is sellers selling at a relative prie below their opportunity ost. Another type of error is buyers buying at a prie above their opportunity ost. The olumn labeled either or both omputes the perentage of transations in whih at least one of the parties, seller or buyer, made an error. Finally, as explained in Setion II, the Riardian model makes lear preditions about the diretion of trade. Beause all players in the same ountry have the same opportunity ost, they an t make gains from trade by trading with eah other. Also, the model predits that ountries will sell the good in whih the ountry has a omparative advantage; trade between two ountries in the opposite diretion will also entails losses for at least one of the parties. The olumn labeled wrong ountry omputes perentages of transations in a diretion other than predited by the theory. The data shows a relatively large perentage of errors in transations: 44% in the wine market and 51% in the loth market. Most of these errors are assoiated with transations in a diretion other than predited by the theory. Interestingly, in the wine market, whih is the one where we found a greater degree of prie onvergene, the perentage of errors ( either or both olumn) delines from round to round, showing an evidene of learning. However, in the loth market, errors remain at about 50% of the transations in all the rounds.
4/5/2005 The basis of international trade: A lassroom experiment page 12 To better understand the large perentage of errors, we ompiled data on trade errors for eah individual player, where we found that most of the errors were onentrated in a relatively small number of players. Of 23 players, the 3 players with most trading errors aounted for 68% of all the selling errors and 49% of all the buying errors. On the other extreme, 48% of players made no selling errors and 26% of players made no buying errors. The rest of the players, while making an oasional error, gained from the majority of their trades. The presene of errors may be due in part to the fast pae of the game and the diffiulty in analyzing numerial data under time pressure. We expet that the new graphial interfae under development will give players a muh more intuitive way to evaluate how they are doing and make their deision-making easier. In any ase, students trading errors gives instrutors a golden opportunity to motivate students to improve their understanding of the Riardian model. A lass disussion based on game results on prodution and onsumption deisions, market pries, and trading errors ould be very insightful. Although we have not disussed results in lass this time, we believe that that doing so ould greatly enhane the pedagogial potential of the game. The disussion an ertainly go beyond omparative advantage and the Riardian model of trade. Issues of funtioning of markets, noise traders, and trading strategies ould naturally arise when disussing the results of a Riardian Explorer game, motivating students to learn more eonomis. V. Evaluation of Teahing Effetiveness In this setion, we provide the results of teahing effetiveness evaluation onduted in the Fall 2004 semester using 2 setions of Introdution to Eonomi Theory (priniples) ourse at Wesleyan taught by the same instrutor. This ourse is a gateway ourse to the eonomis major at Wesleyan and is a alulus-based one semester introdution to miro- and maro-eonomis. One setion (Setion 2) of the ourse (seleted using a flip of a oin) played the game prior to
4/5/2005 The basis of international trade: A lassroom experiment page 13 attending a leture on international trade. Both setions were exposed to the same leture, homework assignment and additional readings. The ontrol setion (Setion 1) partiipated in an unrelated experiment that studied gender differenes in performane. Both setions took the same final exam. There were 53 students in Setion 1 and 50 students in Setion 2, out of whom 33 played the game. There was one multiple hoie question on the final that dealt with the Riardian model. We evaluate teahing effetiveness by looking at how likely students were to answer this question orretly. The question provided unit labor requirements and total available labor for a 2-ountry 2-good Riardian model 7. In the orret solution, students had to realize that due to parameter hoie while one ountry had an absolute advantage in the prodution of both goods, Riardian model predited full speialization by one ountry only, while the other ountry produed both goods. Simply applying omparative advantage formula to deide on the pattern of speialization did not produe a orret answer. 76% of students in the ontrol setion blindly applied the omparative advantage formula and only 14% of students got the question right. In ontrast, 51% of students in the treated setion used omparative advantage formula and 31% answered the question orretly. It is worthwhile to mention that setion 1 was overall a stronger setion with students onsistently getting higher sores on all exams. For further support, we ran several speifiations of a linear probability regression to see how likely students were to answer the final question orretly depending on whether they partiipated in the game. Results of this analysis are reported in Table 2. Similar results an be obtained by running probit and are available upon request. In all speifiations, the dependent variable is 1 if a student answered the final exam question orretly and 0 otherwise. Speifiation (1) only uses partiipation in the Riardian game as an explanatory variable. Column (2) in addition ontrols for individual sores on the first and seond midterm exams as a proxy for student ability. Columns (3) and (4) add sores on trade homework assignments to speifiations (1) and (2), respetively. This dereases the
4/5/2005 The basis of international trade: A lassroom experiment page 14 sample size as the problem sets did not ount for students grades and therefore not everybody ompleted them arefully, if at all. The seond problem set had harder questions. None of the questions on the problem set were similar to the final exam question. First we look at the pooled regressions, ombining both setions of the ourse. As expeted, performane on prior exams strongly predits the likelihood of answering the trade question on the final orretly. Depending on the speifiation, the probability of answering the question orretly is between 32% and 43% higher for students who played the game. These results are statistially signifiant at the 1% level. Similar results an be obtained by looking only at the treated setion and ontrolling for the fat that not every student partiipated. These results are meaningful beause nonpartiipation (exept in one ase) was not a result of hoosing not to ome to lass but rather due to an already fixed omputer glith that did not allow some students to login. While sample size dereases substantially if we look at setion 2 only, students who played the game are 31% to 46% more likely to answer the final question orretly. This evidene is very strong and in the future we would like to provide similar evaluations for using Riardian Explorer at other levels of instrution and also with longer and more rigorous tests. In the meantime, we were very enouraged to see these preliminary results. In addition, we onduted surveys at the end of the Fall 2004 and at the beginning of the Spring 2005 semesters to ask students about their experiene with the software. Notie that while we ran the Fall 2004 survey during the exam week, between 1 and 2 months after the students played the game, students answered the Spring 2005 survey right after playing the game. Some interesting evidene from this survey is reported in Table 3. Students answered questions presented on a 5-point sale ranging from 1 (strongly disagree) to 5 (strongly agree). Survey results show that students got better at understanding the game environment after repetition. Therefore it is advisable to play the game more than one during the ourse. Most of them agreed
4/5/2005 The basis of international trade: A lassroom experiment page 15 with the eduational merits and objetives of the software. International trade students overwhelmingly ommented on the benefits of playing the game more than one. VI. Conlusion Experiments are beoming a prominent teahing tool in eonomis. The interative nature of lassroom experiments allows students experiene eonomi onepts as ative partiipants. Furthermore, instrutors an failitate disussion using the Sorati method after the experiment omplementing the traditional, leture-based delivery of the material. Our software provides an easy way to adopt experiments to teah the Riardian model of trade. By failitating student understanding of the fundamental onept of omparative advantage and providing glimpses on deeper eonomi issues, suh as the role of pries in oordinating individual deisions and general equilibrium, the Riardian Explorer is a useful and pedagogially effetive tool in international trade and miroeonomis ourses.
4/5/2005 The basis of international trade: A lassroom experiment page 16 Referenes Anderson L. R., E. Blanhard, K. Chaston, C. Holt, L. Razzolini and R. Singleton. 2001. Prodution and gains from trade. NSF Workshop on Classroom Experiments Working Paper. Ball, S. B. and C. C. Ekel. 2004. Using tehnology to failitate ative learning in eonomis through experiments. Soial Siene Computer Review 22 (Winter): 469-78. Bergstrom, T. C. and J. H. Miller. 2000. Experiments with eonomi priniples: Miroeonomis. Chiago, IL: Irwin MGraw-Hill Publishers. Bhagwati, J. N. 2002. Free trade today. Prineton, NJ: Prineton University Press. Holt, C. A. 2005. Markets, games, and strategi behavior: Reipes for interative learning, <http://www.people.virginia.edu/~ah2k/expbooknsf.pdf>, University of Virginia, aessed 2005. Holt, C. A. 1999. Teahing eonomis with lassroom experiments: A symposium. Southern Eonomis Journal 65 (January): 603-610. Krugman, P. R and M. Obstfeld. 2003. International eonomis: Theory and pratie. 6 th edition. New York, NY: Addison-Wesley Publishers. Krugman, P. R. 1996. Riardo s diffiult idea. <http://www.pkarhive.org/>, aessed 2005. Krugman, P. R. 1993. What do undergrads need to know about trade. Amerian Eonomi Review 83 (May): 23-26. Noussair, C. N., C. R. Plott, and R. G. Riezman. 1995. An experimental investigation of the patterns of international trade. Amerian Eonomi Review 83 (June): 462-491. Porter, T. S., T. M. Riley, and R. L. Ruffer. 2004. A review of the use of simulations in teahing eonomis. Soial Siene Computer Review 22 (Winter): 426-43. Santos, J. 2002. Developing and implementing and Internet-based finanial system simulation game. Journal of Eonomi Eduation 33 (Winter): 31-40. Shmidt, P. J. 2003. Ative and ooperative learning using web-based simulations. Journal of Eonomi Eduation 33 (Spring): 151-67. Woltjer, G. 2005. Deisions and maroeonomis: Development and implementation of a simulation game. Forthoming Journal of Eonomi Eduation.
4/5/2005 The basis of international trade: A lassroom experiment page 17 Figure 1: Consumption Possibility Frontier for Prodution Choie ( p p y x, ) PPF CPF x y p x p y LY i a L Y i p V X i p V LX i a L
4/5/2005 The basis of international trade: A lassroom experiment page 18 Figure 2: Game Sreen
4/5/2005 The basis of international trade: A lassroom experiment page 19 Figure 3: Reahing the PPF in Portugal (Riardos game, Fall 2004, rounds 2 & 4) Figure 4: Gains from trade in Portugal (Riardos game, Fall 2004, rounds 2 & 4)
4/5/2005 The basis of international trade: A lassroom experiment page 20 Figure 5: Relative pries (Riardos game, Fall 2004) Pw/P 1.67 1.50 1.33 1.17 1.00 0.83 0.67 0.50 0.33 0.17 0.00 Wine Market Transation P/Pw 10.5 9 7.5 6 4.5 3 1.5 0 Cloth Market Transation Round Sellers sell too low Table 1: Trading Errors (Riardos game, Fall 2004) Wine Market Buyers buy too high Either or both Wrong ountry Sellers sell too low Cloth Market Buyers buy too high Either or both Wrong ountry 1 6% 61% 67% 50% 44% 17% 52% 48% 2 29% 17% 46% 42% 23% 28% 50% 50% 3 25% 21% 38% 38% 26% 26% 53% 50% 4 13% 22% 30% 26% 21% 33% 51% 39% Mean 19% 28% 44% 38% 27% 27% 51% 46%
4/5/2005 The basis of international trade: A lassroom experiment page 21 Table 2: Teahing Effetiveness of the Riardian Explorer game A. Setions 1 and 2 Pooled (1) (2) (3) (4) Played Riardian Explorer? (1=yes) 0.323*** 0.355*** 0.373*** 0.432*** (0.094) (0.082) (0.110) (0.098) Sore in First Midterm Test 0.945*** 0.789** (0.262) (0.310) Sore in Seond Midterm Test 0.358* 0.670** (0.185) (0.308) Sore in First Trade Problem Set -0.127-0.146 (0.190) (0.214) Sore in Seond Trade Problem Set 0.584** 0.463** (0.237) (0.214) Constant 0.101*** - 0.655*** -0.237-1.046*** (0.036) (0.143) (0.224) (0.312) N. Observations 103 103 79 79 R-squared 0.14 0.34 0.23 0.41 B. Setion 2 Only (1) (2) (3) (4) Played Riardian Explorer? (1=yes) 0.424*** 0.316*** 0.461*** 0.433*** (0.088) (0.087) (0.115) (0.148) Sore in First Midterm Test 0.916** 0.606 (0.354) (0.720) Sore in Seond Midterm Test 0.392 0.866 (0.247) (0.661) Sore in First Trade Problem Set -0.207-0.210 (0.325) (0.278) Sore in Seond Trade Problem Set 0.284 0.252 (0.427) (0.401) Constant 0.000*** - 0.623*** -0.035-0.864* (0.000) (0.172) (0.224) (0.489) N. Observations 50 50 32 32 R-squared 0.20 0.36 0.19 0.37 Notes: Robust standard errors in parentheses. ***, ** and * indiate signifiane at the 1%, 5% and 10% onfidene levels, respetively.
4/5/2005 The basis of international trade: A lassroom experiment page 22 In the first round of the game my prodution and trading deisions were almost random. In the last round of the game my prodution and trading deisions were almost random. The game helped me understand opportunity osts. The game helped me understand omparative advantage. The game helped me appreiate even more the role pries play in oordinating eonomi ativity. It would be more useful to play the game more Table 3: Surveys of Partiipants International Trade Fall 04 0.43 (0.36) Introdutory Eonomis Fall 04 0.47 (0.47) Introdutory Eonomis Spring 05 0.50 (0.29) Experimental Eonomis Spring 05 0.52 (0.36) 0.00 0.00 0.13 0.04 (0.93) *** (0.94) *** (0.75) *** (0.84) *** 0.64 0.39 0.77 0.44 (0.07) *** (0.11) * (0.09) *** (0.28) 0.77 0.58 0.79 0.48 (0.00) *** (0.11) *** (0.08) *** (0.20) ** 0.92 0.72 0.70 0.64 (0.00) *** (0.11) *** (0.09) *** (0.20) *** 0.93 - - (0.00) *** than one Number of responses 14 19 24 25 Partiipation rate 61% 59% 100% 100% Notes: Bold numbers are the proportions of agree (4) and strongly agree (5) responses (p A ). Number in parentheses are the proportions of disagree (2) or strongly disagree (1) responses (p D ). We tested for the equality of the proportions of agreements and disagreements (H 0 : p A = p D, H 1 : p A p D ). ***, ** and * indiate that the test is signifiant, respetively, at the 1%, 5% or 10% onfidene levels.
4/5/2005 The basis of international trade: A lassroom experiment page 23 Endnotes 1 The software is available at http://www.wesleyan.edu/re. 2 Paper-and-penil lassroom experiments to teah omparative advantage an be found in Anderson et al. (2001) and in Bergstrom and Miller (1997). A researh experiment on the subjet using a more ompliated environment is in Noissair, Plott, and Riezman (1995). 3 Holt (1999, 2005) provides a omprehensive analysis of the history and pratie of employing experiments for instrutional purposes. See also reent reviews by Porter, Riley, and Ruffer (2004) and Ball and Ekel (2004). Reent appliations inlude Santos (2002), Shmidt (2003), and Woltjer (2005). 4 Refer to Setion II for details on the model on whih the game is based. 5 In our experiene, using a ρ = -60 provides a very good approximation to the Leontieff utility. 6 We have usually set up games of 4 rounds. 7 Consider 2 Island eonomies: Crete and Rhodes. Inhabitants produe 2 goods, fish and bread. Unit labor requirements for prodution of eah good are in the table below. Inhabitants of Crete and Rhodes like to onsume both goods in fixed proportions. Available labor: Crete 600, Rhodes 400. Crete Rhodes Fish 2 6 Bread 3 4 (a) Rhodes has an absolute advantage in the prodution of Bread. (b) Crete has a omparative advantage in the prodution of Bread. () Aording to the Riardian model, when islands trade, Crete will speialize in Fish and Rhodes will speialize in Bread. (d) Aording to the Riardian model, when islands trade, Crete will produe both Fish and Bread, but Rhodes will produe Bread only. (e) Aording to the Riardian model, when ountries trade, Crete will produe only Fish, but Rhodes will produe both Fish and Bread.