Rural-Urban Migration, Urban Employment and Underemployment, and Job Search Activity in LDCs

Transcription

1 Cornell University ILR School Research Stdies and Reports ILR Collection Rral-Urban Migration, Urban Employment and Underemployment, and Job Search Activity in LDCs Gary S. Fields Cornell University, Follow this and additional works at: Part of the Growth and Development Commons, Income Distribtion Commons, International and Comparative Labor Relations Commons, International Economics Commons, Labor Economics Commons, and the Regional Economics Commons Thank yo for downloading an article from Spport this valable resorce today! This Article is broght to yo for free and open access by the ILR Collection at It has been accepted for inclsion in Research Stdies and Reports by an athorized administrator of For more information, please contact

2 Rral-Urban Migration, Urban Employment and Underemployment, and Job Search Activity in LDCs Abstract [Excerpt] In this paper, we shall present a formal theoretical mode with which to analyze the eqilibrim allocation of the labor force between labor markets. Or basic premise is that the same kinds of forces that explain the choices of workers between the rral and rban sectors can also explain their choices between one labor market and another within an rban area and are probably made simltaneosly. The decision-makers -- be they individals or family nits are presmed to consider the varios labor market opportnities available to them and to choose the one which maximizes their expected ftre income. Keywords labor markets, income, rban migration, nemployment, less developed contries Disciplines Growth and Development Income Distribtion International and Comparative Labor Relations International Economics Labor Economics Regional Economics Comments Sggested Citation Fields, G. S. (1972). Rral-rban migration, rban employment and nderemployment, and job search activity in LDCs [Electronic version]. Washington, DC: U.S. Agency for International Development. Reqired Pblisher Statement A later version of this report was pblished as: Fields, G. S. (1975). Rral-rban migration, rban employment and nderemployment, and job search activity in LDCs. Jornal of Development Economics, 2(2), doi: / (75) This article is available at DigitalCommons@ILR:

3 RURAL-URBAN MIGRATION, URBAN UNEMPLOYMENT AND UNDEREMPLOYMENT, AND JOB SEARCH ACTIVITY IN LDCs Cary S. Fields December, 1972 In recent years, the rban areas in less developed contries have grown very rapidly. Between 1950 and 1960, rban areas in Africa grew by 69%, in Latin America by 67%, and in Asia by 51%, while rral areas grew by only 20% over the same period. Since biological growth rates rarely exceed 3% per annm, mch of the rban growth is de to rral-rban migration. There is a growing consenss on a nmber of aspects of the migration qes 3 tion. Both economists 2 and non-economists agree that rral-rban migration can be explained primarily by economic factors: the "psh" from agricltre and the "pll" of relatively high rban wages. The "bright lights of the city" and other cltral explanations are given relatively little weight in the literatre. There is also agreement that sch migration is qite rational despite the existence of rban nemployment. The essence of this relationship is smmarized clearly in perhaps the best-:mown article on the sbject, that of Harris and Todaro: "... migration proceeds in response to rban-rral differences in expected earnings (defined below) with the rban employment rate acting as an eqilibrating force on sch migration." 4 Finally, it is agreed that yong persons are most likely to migrate 5 and that they experience mch higher rates of rban nemployment than other workers. 6 The reasons for the

4 -2 greater propensity of the yong to move to economicrily advantageos areas have been discssed by a nmber of writers 7 and will not be repeated here. In this paper, we shall present a formal theoretical mode with which to analyze the eqilibrim allocation of the labor force between labor marketg. Or basic premise is that the same kinds of forces that explain the choices of workers between the rral and rban sectors can also explain their choices between one labor market and another within an rban area and are probably made simltaneosly. The decision-makers -- be they individals or family nits are presmed to consider the varios labor market opportnities available to them and to choose the one which maximizes their expected ftre income. Or point of departre is the received theory of rral-rban migration in less developed contries, which is the model of Harris and Todaro (1970). We begin by smmarizing the basic featres of the model. While we accept the basic approach, we show that the particlar implication of the model with respect to the eqilibrim rban nemployment rate sbstantially overstates the rates actally observed by Trnham (1970) and others. We then extend the analysis to take into accont a nmber of important factors which have previosly been neglected--a more generalized approach to the job search process, the possibility of nderemployment in the so-called rban "mrky sector," preferential treatment by employers of the better-edcated, and consideration of labor trnover--and demonstrate that the reslting framework gives predictions closer to actal experience.

5 -3 1. The Received Theory of Rral-Urban Migration The received theory of rral-rban migration, first set forth in Todaro (1968), has been revised and agmented by Todaro (1969), Harris and Todaro (1970), again by Todaro (1971), and by Johnson (1971). The Rarris-Todaro version is best known and we shall consider it in that form. The '.iodel treats rral-rban migration primarily as an economic phenomenon. In essence, the theory postlates that workers compare the expected incomes in the rban sector with agricltral wage rates and migrate if the former exceeds the latter. Rral-rban migration is ths the eqilibrating force which eqates rral and rban expected incomes and as sch is a diseqilibrim phenomenon. The three basic characteristics of their model--tha migration occrs largely for economic reasons, that the migration decisio depends on expected rather than nominal wage differentials, and that migration takes place in diseqilibrim--sggest that rral-rban migration be given a new emphasis. Rather than considering it as a key phenomenon in its own right, migration cold better be regarded as the adjstment mechanism by which workers allocate themselves between different labor markets, some of which are located in rban areas and some in rral areas. Harris and Todaro formlate the problem in the following way. Let W a and W respectively denote the nominal agricltral and rban wage rates, E be the nmber of rban jobs, and L be the rban labor force. The expected rban (i ~ - w )

6 -4- Expected rral income (E(Wa)) is simply Wa. The amont of rral-rban migration (L) is a fnction of the rban-rral expected wage differential ( ) - ( (W) - (a The rral-rban eqilibrim condition (3) E(W) E(Wa ) becomes (4) W E = W 8 L- a and the eqilibrim employment rate is (5) E W4 L W How does this prediction sqare with available empirical evidence? Not well. Per capita incomes in rban areas are anywhere from two to eight times as high as in rral areas. 9 Ths, Harris-Todaro wold predict rban employment rates of 1/2 to 1/8. Yet the highest nemployment rate observed in seventeen less developed contries is 20%. 1 0 While it might be argied that eqilibrim has not yet been reached, it seems mch more likely in light of the size of the gap between actal and predicted nemployment rates tfat the theory as stated needs to be amended to conform more closely to the observed facts. This is or task in the remaining sections.

7 2. A More Generalized Formlation of the Job Search Procese In the Harris-Todaro model, the probability of obtaining an rban job is defined as the nmber of rban jobs divided by the rban labor force. -5 Implicitly, this specification assmes that persons living itn rral areas hav3 no chance whatever of finding rban jobs. In this section, we shall show that the Harris-Todaro specification implies a higher eqilibrim nemployment rate than wold be predictdd by a more generalized formlation of the job search process. There are several reasons why rral residents wold ba expected to have a positive chance of obtaining rban jobs. Mch rban hiring is done throgh channels which do not exclde rral residents. Some jobs are "advertised" and filled informally by word of moth. An rban resident may locate a job for a friend or relative and then send word (and money) for him to come to the city. Other jobs are filled by a central labor exchange with which rral residents are able to register. Finally, those persons in rral areas proximite to cities may on occasion be able to look actively for an rban job. However, a nmber of factors make it probable that locating in the cities and searching for a job wold still have a positive payoff. These inclde delays in conveying information to persons in rral areas, the preference of employers for personal contact with prospective employees, the costs of repeated visits to cities in search of work, and the simple fact that many jobs are fond by happening to be at the right place at the right time. All these cotsiderations may be smmarized by a single parameter. Shold an rban job become available, each rban resident wold have some particlar chance of being selected for it and each rral resident wold have some lesser chance. Let the relative chance of any given rral worker obtaining the job relative to any given rban worker be denoted by n. We shall call this nmber n

8 -6 the relative rral-rban Job search-parameter. It is an (inverse) index of the payoff to job search; when job search is profitable, n is low, and vice versa. To give an example, sppose n n 1/2 and let the probability that a worker who resides in the rban sector will obtain an rban job in the crrent period be 0.8. Then, the probability that a comparable worker who lives in the rral sector will obtain an rban job is 0.4. Similarly, if n were eqal to 1/4, the probability of a rral worker secring an rban job wold be 0.2. The vale of the rral-rban job-search parameter n In a given contry may be presmed to depend on a nmber of economic and cltral variables inclding the lengtn of the work week in agricltre, the extent of favoritism, nepotism, and discrimination in the labor market, and the efficiency of the labor exchange. We wold expect that other things eqal, the longer the work week, the smaller the job-search parameter n. This is becase a longer work week leaves fewer hors for other activities inclding job search. Ths, the longer the work week, the poorer the relative chance of rral workers obtaining rban employment and so the lower the rral-rban job-search parameter. The greater the degree of nepotism, favoritism, and discrimination in an important, economy, the greater the expected job-search parameter. When these factors are one's contacts, skin color, tribal origin, or other personal characteristics have a greater bearing on his employment stats than the extent of his job search. Members of the favored grop cold remain on the farm and jst wait to be called; their prospects wold be little improved by migrating to the city and searching fll time; Persons not in the favored grop wold

9 -7 have almost as poor a chnce of obtaining a job from the farm as they wold in the city where they are discriminated against; their prospects also wold be little improved by fll-time -ob-search in the city. Conseqently, a high jobsearch parameter wold be expected where favoritism and discrimination are prevalent. Finally, an efficient labor exchange, in which most rban job openings are filled by lotteries condcted by the labor exchange, wold case there to be little payoff to job serarch and raise the Jcb search coefficient to near nity. For instance, this was the case in Kenya dring the Tripartite Agreement of 1970, at which time all employers were reqired to increase employment by ten percent, workers were reqired to register with the lcbor exchange, and the lottery reslts were pblished in the daily press. Let s now incorporate the generalized job-search formlation into the Harris-Todaro model and determine the reslting effect on the eqilibrin nemployment rate. We shall denote the probability of a given rban resident becoming employed in an rban job by P' Assming that all jobs are available to all persons eqally, 11 (6) P F U J where E is rban employment and J is the nmber of Job-seeker eqivalents, defir 3d as follows. J is a weighted sm of the rban and rral labor U forces, the weights reflecting the relative chancec of being hired. Since each rral resident has only an n'th as great a chance of being hired, a weight of one is assigned to each rban resident and a weight of n to each rral resident.

10 -8- Letting L and L a be the nmber of residents in the rban and rral areas respectively, we therefore have, by definition, (7) J L + nl. U a J is called the nmber of job-seeker eqivalents becase the same nmber of rban residents as J and no rral residents wold lea'e each with an eqivalent probability of finding rban employment. The labor market jst described may be thoght of as fnctioning like a lottery in which each rban resident (L ) has one ticket, each rral resident (L a ) has an n'th of a ticket, each ticket is identical, each prize is eqally valable, and the total nmber of pa!zes is E. The expected wage of a member of the rban labor force (E(W )) is simply the rban wage W times the probability of employment P : U (8) E(W) W P U W U U 7 The expected income of a rral resident is slightly more involved, since it depends on whether or not he is hired for an rban job. If he does obtain an rban job, he will earn the rban wage W ; otberwise he earns the rral wage W a - The respective probabilities are n E E and 1 - n. Therefore, the expected 3 J income of a member of the rral laboruforce is (9) E(W) W 1 + W a ( - n E ). g

11 -9- becomes The rral-rban migration eqilibrim condition E(W ) E(W ) U a E E ~w (~ E (10) W Ev. - W n U T U T E + W (1 - n a The sitation described by the Harris-Todaro model is easily seen to be the case where n - 0. In order to determine the eqilibrim employment rate, we solve the eqilibrim condition (10) for E. Sbstitting (7) for J and (L - L ) for Lat L where L is the total labor force, we find (11) 1 + n (L - 1) EUE L ih: WIT ~~_ W a W a E 1- oeewhnarcla In the Harris-Todaro case where n - 0, E_ W 1 However, when agricltral EL w a 1 workers have some chance of obtaining rban employment and n > 0, L > W Wa Frthermore, by differentiating (11) with respect to n, one can easily see that the larger is n, the larger is the rban employment rate. Ths we find U that there is a lower eqilibrim nemployment rate in general than -7old be predicted by the Harris-Todaro model and the greater the relative chance of rral workers finding rban jobs, the greater the discrepancy between the general reslt and the Harris-Todaro reslt.

12 The Introdction of a Mrky Sector The natre of the migration process and the reslting rban growth are perhaps best described impressionistically by a typical scenario. New arrivals in the cities ordinarily stay with friends or relatives who help hose and feed them while they look for work. A dozen or more people crowded into one room is not ncommon. They need not live in hosing which is rented or provided as part of job compensation. Sqatter settlements and shanty towns hose a sbstantial portion of rban poplations, particlarly in Africa. Unemployment (by standard definitions) is not very common. Additional hosehold members are expected to contribte to their sp-irt. Freqently, they assist with the hosehold chores by preparing meals, washing clothes, or caring for children. Simltaneosly, they search for work (albeit on an irreglar basis) and are classified as nemployed. The most fortmnate new migrants obtain a permanent modern sector job as a clerk, messenger, or whatever. However, these are the best jobs and the typical migrant is forced to find some lesser means of earning a cash income. le may secre one or more typically a sccession of wage jobs (e.g., hose-servant, cook in a small lnch kiosk, assistant in a family shop)or engage In self-employment (e.g., selling prodce, newspapers, crios, or shoe shines on the street corner). These activities have been given several names inclding petty capitalism, the traditional sector, the service sector, and the grey area. A particlarly graphic term, and the one we shall se to denote this whole range of activities, is the ":mrky sector."

13 -11- Entry into the mrky sector is typically open. For instance, a person can get started by bying some peas in the market, removing the pods at the side of the road, and selling podded peas to passers-by a- a higher price. Prostittion is another occpation which has notoriosly easy -ry. Workers in the mrky sector are ordinarily classified as employed 1 2 althogh they themselves and the statisticians who measrn those things wold be inclined to consider them nderemployed. An examination of available time series evidence sggests that nemployment rates have not in general worsened sbstantially over time. Of ten contries which permit analysis, nemployment rates have risen 1 3 in three (Korea, Colombia, and Panam), fallen in five (UAR, Taiwan, Argentina, Chile, and Perto Rico), and remained nchanged in two (Philippines and Trinidad-Tobago).14 The tentative conclsion to be drawn is that most migrants have encontered limited sccess and are engaged in some sort of mrky sector employment. The existence of opportnities for paid employment in the mrky Rector gives each member of the labor force a new option. Not only can he choose between staying in (or retrning to) agricltre or being either employed or nemployed in the cities, bt he can algo volntarily choose to be nderemployed in the rban mrky sector while looking for a better job. Why don't all workers enter the mrky sector? While nderemployment in the mrky sector yields a positive wage and nemployment pays no wage, the mrky sector income is earned at the cost of redced job search opportnities. This may be simply becase mrky sector workers have less time to look for

14 -12 modern sector jobs or for some other reason. In the remainder of this section, we shall examine these effects and show that introdction of the mrky sector leads to a lower eqilibrim nemployment rate than the Harris-Todaro reslt. The mrky sector may be introdced in a manner similar to the development of the agricltral sector in the last section. In order to keep the effects of recognizing the mrky sector separate, we retrn to the original model (eqations (1) - (5)) and assme that agricltral workers have no chance of obtaining modern sector jobs. We now have two kinds of rban jobs, modern sector and mrky sector, with wage rates W and W m respectively. While we will hold W constant as before, we will regard W m as an endogenos variable to be determined by the model. Let the relative job search parameter between mrky and mcdern sector jobs he denoted by h. The parameter h is the probability that any given person in the mrky sector labor force wold be hired for a modern sector job relative to the probability of any given member of the modern sector labor force being hired. Since we are now assming that rral residents have no chance of obtaining rban jobs, the nmber of job-seeker eqivalents for modern sector rban jobs is (12) J L + hl, where L and L are the modern sector and tnrky sector labor forces respectively. U m Eqilibrim between the mrky and modern sectors reqires that the expczted wage in the modern sector (E(W)) eqal the expected wage in the mrky

15 sector (E(Wm)). By analogy with the expected agricltral wage nder general job sear:h conditions, As before, (13) E(W m )= W h E + ( - h EW 7- J m U (14) E(W) E. J Eqilibrim between the rral and rban sectors reqires that these in trt, eqal the expected agricltral wage (E(Wa)), which is (15) E(Wa). Was since we are once again assming that agricltral workers have no opportnity of obtaining an rban job. Therefore, the rral-rban and intra-rban eqilibrim conditions may be combined as (16) W a - W h E + (1 - E h ) W = E W m U We now wish to solve for the eqilibrim labor force allocation, rban nemployment rate, and mrky sector wage rate. From the eqality between the first and third members of (16), in eqilibrim, (17) j WE E a Sbstitting this into the eqality between the second and third members and solving for W, we obtain for the eqilibrim mrky sector wage (18) Wa (1-h) m 1 - h a which is constant for particlar vales of W W, and h. a,

16 -14- For the determination of the mrky sector labor force, we assme that the demand for mrky sector otpt (Q.) is the sm of the demand by employed modern sector workers f(e ) pls the demand by mrky sector workers ilm): m (19) Qm - f(e) + g(lm), f'>o, g'>o. We frther assme that since mrky sector workers are nderemployed, the demand is spplied an4 the resltant income is shared eqally among mrky sector workers, i.e., (20) W. -m m Sbstitting (18) and (19) into (20), we obtain an implicit expression for the mrky sector labor force as (21) [fve I + g(lm)j - W (i-h) L m W 1-h a W In this general form, we cannot solve explicitly for L m. However, if we adopt the simplifying assmption that the amon. of mrky sector otpt demanded by mrky sector workers is fixed, this along with the assmed constancy of E implies that the total mrky sector otpt is fixed at some level Qm. Sbstitting im for [f(e) + g(lm)i in (21) and rearranging, we derive the mrky sector labor force as (22) L - m (1-h Wa) W Wa a (1-h)

17 -15- Sbstitting (22) and (17) into (12) and rearranging, we find that the total rban labor force(lr b ) is W WE a The rban employment rate is modern sector employment pls mrky sector employment divided by the total rban poplation: E +L (24) m Lrb Sbstitting (22) and (23) into (24), we find that the rban employment rate is + (25) Wa (1-h) WE + =. W a where U. = n (1-h Wa ). W U This may readily be shown to be greater than Harris- observing that the reslt is nambigosyy positive. We have therefore demonstrated the validity of the proposition that introdction of the mrky sector leads to a lower eqilibrim nemployment rate than predicted by the Harris- Todaro model. What is the effect of the size of the mrky-modern relative job search parameter on the eqilibrim rban employment rate and other labor market variables? Or model sggests that the greater the chance of a worker employed

18 -16 in the mrky sector of obtaining s modern sector job relative to an nemployed worker who searches fll-time. (i.e., the larger is h): a) the smaller the eqilibrim mrky sector wage rate, b) the larger the eqilibrim mrky sector labor force, c) the smaller the modern sector labor force in eqilibrim, d) the smaller the total rban labor force in eqilibrim, and e) the larger the eqilibrim rlan employment rate. Point a) may be demonstrated by partially differentiating (18) with respect to h W (26) am M a (Wal (1-hWa) 2 W and noting that the reslt is negative since W U > W a (22) To show b), differentiate W LW (l1 )J2 (27 a m Wo e)u1)ad(7 and observe that the reslt is positive for W > W give ;i a For c), (12) end (17) (28) L= WE -hl W a which clearly varies inversely with h. Part d) is easily seen from the expression for the eqilibrim rban labor force in (23).

19 -17- Finally, differentiation of (16) with respect to h gives a constant J and ()w m (W - W ) E (30) -, = % < 0. i-hr- U 3L (20) and (30) imply - > 0, which along with the constancy of J implies U h that We now have more rban residents employed In the mrky sector and fewer nemployed seekine modern sector nskilled jcbs and therefore an nambigosly higher rban employment rate for a larger vale of h. A priori considerations sggest that the mrky sector relative job search Darameter h wold be fairly large, i.e., worker' job search activity wold not be seriosly impeded by taking a mrky sector job rather than remaining nemployed in search of work in the modern sector. This wold seem so for two reasons. First, the natre of the mrky sector is sch that self-employment, flexible hors, and part-time work are common. Ths, it is often possible to adapt one's work week and the specific work hors so as to be relatively free to search for modern sector jobs. Second, many modern sector jobs are obtained by contacts from employed friends or relatives. Conseqently, workers wold have relatively little to gain by searching fll time and they wold be more likely to take p employment in the mrky sector in order to earn a cash Income. To the extent that these two considerations hold, rban nemployment rates are likely to be fairly low in absolte terms as well as relative to the prediction of the Harris-Todaro model. However, it shold not be forgotten that these low nemployment rates conceal a considerable volme of nderemployment in the mrky sector.

20 -18- As long as the mrky sector labor force has some positive chance of becoming employed in the modern sector, the eqilibrim mrky sector wage wold 1 5 be less than the agricltral wage. This wold be expected becase the lower wage is the price workers mst pay in eqilibrim in order to have a better chance of obtaining a relatively high-payinr modern sector job. This gives an additional reason for the existence of an impoverished rban class. Not only are some people willing to be nemployed mch of the time in order to earn high wages when they are employed in the modern sector, bt others are willing to be nderemployed by working for very low wages (less even than they cold earn in agricltre) in order to have a better chance of being hired for those same modern sector jobs. 4. Preferential Treatment by Employers of the Better Edcated A nmber of observers of lees developed contries report employers sing edcational attainment as a criterion for hiring and selecting the better edcated in preference to those with less edcation. 1 6 What effect does this have on the eqilibrim employment rate? Let s once again retrn to the original model and neglect the possibility of employment in the mrky sector. Now sppose there are two categories of workers: the edcated (L ) and nedcated (L), of whom L live in rban areas e and La in agricltre. Sppose frther that becase of this systematic preference by employers the available spply of edcated workers are hired immediately withot nemployment and nedcated workers mst divide the remaining jobs.

21 The expected income of art nedcated worker who enters the rban labor force (E(WI )) is E -L (31) E(WI ) - W -el and the expected income of an nedcated wo'!er who enters the agricltral labor force (E(WI a) is (32) E(WI a) - was Eqilibrim between the two labor markets for nedcated workers reqires that E(WI ) - E(W a) or E-L (33) W e. L a The eqilibrim employment rate for nedcated workers is E-L (3) e, L -a W_ one. The total and the eqilibrim employment rate for edcated workers is employment rate is a weighted average of these two rates, the weights given by the percentage of nedcated and edcated workers respectively. Therefore, the total rban employment rate in eqilibrim is (35) L W + Le Lr W L rb rb which is clearly greater than the llarris-todaro reslt ( a) except when 1, 0. e

22 -20- The reason for this greater employment rate is inherent in the job search mechanism itself. When an edcated worker is hired, he fills a position which some greater nmber of nedcated workers had been seeking. For example, if a W, the eqilibrim employment rate for nedcated workers is I as given a 3 3 by (34). For each edcated worker who is hired preferentially, there is one less rban job available to nedcated workers and in eqilibrim there wold be three fewer job seekers. 5. Consideration of Labor Trnovei: The basic Harris-Todaro model has been extended by Johnson (1971) to 17 Pive explicit attention to a time dimension and rate of labor trnover. As before in choosing between the rban and rral sectors workers are assmed to consider the expected incomes in each. Now, however, the present vale of the expected lifetime income streams are relevant. For a person in the rban labor force, this is T (36) V of E(W)e - rtdt, where T is his relevant time horizon, E(W ) is the expected rban wage which varies over time, and r is a discont rate. While T and r have clear interpretations, it is not at all obvios what the appropriate vale of E(W ) is. Even if we grant that the expected rba,wage E(W ) at any time t in the "objective" mathematically expected wage, t.e., (37) E(W) a W t where W is the rban wage at time t and 0 the probability of being employed U

23 -21 at that time, the appropriate ftre vales of W and 0 are nonetheless sbjective. In this sense there are as many different expected rban incomes as there are workers with different notions abot ftre wages and employment probabilities. If workers' behavior is standardized and we assme that all workers behave as if today's wage and probability of finding employment will prevail forever, Johnson shows that the expected probability of being employed at any ftre time t is (38) 0- P (1-e' U ) -+0 where 0 is the rate of involntary labor trnover Ln rban jobs. Sbstitting (37) and (38) into (36) and integrating, for a sfficiently yong worker with a long time horizon (i.e., large vale of T), the present vale of expected rban income is W P (39) V U r r+p+* Similarly, if agricltral workers are exclded from consideration for rban jobs bt always Lave the opportnity of earning the agricltral wage Wa the present vale of expected lifetime income in agricltre wold be T W (40) V - f W a ert 0 d a --a. r Rral-rban migration eqilibrim reqires that the expected present vales in each labor force (V and V a ) be eqal. All terms expect F are parameters of the model. Assming randomness of hiring, P (the probability

24 -22 of finding an rban job) is the ratio of hires to Job-ceekers. Johnson shows that the rban nemployment rate varies directly with the rate of labor trnover, provided the individal's discont rate exceeds the rate of growth of rban employment. The Harris-Todaro model. has no job fixity, i.e., infinite labor trnover, and therefore predicts a higher nemployment rate in eqilibrim than wold be expected for any finite rate of labor trnover. 6. Conclsion In this paper, we have soght to nderstand the determination of the eqilibrim level of nemployment in less developed contries. Following the precedent established by!arris and Todaro, we have focssed on the volntary movement of workers between labor markets as the eqilibrating force instead of the more cnventional mechanism of wage adjstment. Within this framework of qantity rather than price adjstment, we have taken into consideration for additional factors: a more generalized accont of the process of search for rban Jobs, the possibility of nderemployment in the so-called mrky sector, the chance that edcaced workers might be favored by employers in job hiring, and recognition of labor trnover in a mltiperiod framework. We have shown that each of these extensions implies a lower eqilibrim nemployment rate than is predicted by Harris and Todaro. Since rban nemployment rates are observed to be mch lower than the Harris-Todaro model predicts, these extensions permit s to retain the qite plasible notion of qantity adjstment as the eqilibrating mechanism in labor markets and yet also have a theory which ls not contradicted by the facts.

25 -23- In interpreting these reslts, one shold not jmp to the conclsion that things are not (or will not become) as bad as the Harris-Todaro model might have led s to believe. It is important that we remember that these nemployment rates fail to take into accont employment at very low wages or the plight of the working poor. Poverty is no less real when people eke ot sbsistence in agricltre or earn less than a living wage while nderemployed in the mrky sectors of the cities. In fact, the social conseqences of a low nemployment rate may be severe, for if planners and policy-makers mistakenly regard nemployment rates of 10-20% as indicating that 80-90% of the rban poplation are flly and gainflly employed, they may fail to act to increase earnings opportnities.

26 -24- FOOTNOTES (1970, p. 1. See Trnham (1970, p. 107). 2. See Harris and Todaro (1970), Stiglitz (1969), and Frank (1971). 3. Especially Cgler ( Harris and Todaro (1970, p. 128). 5. Todaro (1971). in 22 contries, Trnham 6. On the basis of evidence from rban areas workers is doble whole." 47) concldes that "in most cases the rate of nemployment among yong or more than doble that applying to the labor force as a 7. See Sjaastad (1962) and Bowles (1970). 8. In or discssion we shall employ a somewhat different wage determination process from that of Harris and Todaro. The Harris-Todaro model fixes the rban wage rate in real terms. The rral wage is specified as the marginal prodct of labor in agricltre, which depends on the nmber of agricltral workers and the terms of trade between agricltral prodcts and manfactres. Harris and Todaro specified the agricltral wage in this way in order to be able to consider the welfare implications of varios government policies with regard to rral-rban migration in a general eqilibrim framework. Since or present concern is with employment and nderemployment and other labor market conditions, we shall sbseqently treat the rral-rban terms of trade as contained in the rral and rban wages and ignore changes in relative price levels. Frthermore, we will treat the agricltral wage rate as fixed. While this is primarily for expositional prposes, it is also likely that given the small size of the modern rban sector compared to the agricltral sector, the wage e potential migrant cold earn in agricltre wold vary to a relatively small extent over the relevant range and can be treated as constant.

27 Evidence for eight less developed contries (East Pakistan, Egypt, Ivory Coast, Ceylon, Brazil, India, Philippines, and Venezela) is given in Trnham (1970, p. 77). 10. Trnham (1970, p. 57) 11. In sbseqent sections we will distingish between the total nmber of jobs and the jobs for which hiring is taking place and between skilled and nskilled jobs and edcated and nedcated workers. 12. Except for those engaged in Illegal activities which they do not report to censs enmerators. 13. From the first year the series is available ntil the last. 14. Trnham (1970, p. 46) 15. Under more general conditions whereby agricltral workers also have some positive chance of obtaining modern sector employment, any vale of h (the mrky-modern relative job search parameter), greater than n (the rral-rban relative job search parameter) wold pive the same reslt. 16. For evidence on this point, see Blag, Layard, and Woodhall (1969), Kreger (1971), Skorov (1968), and npblished data from the 1971 Nairobf Hosehold Srvey. 17. This extension is described in some detail in my doctoral dissertation. Johnson also extended the model to allow for the possibility that the rban employed might be expected to share part of their incomes with the nemployed, bt this is otside the scope of the present discssion.

28 -26- REFERENCES Blag, H., Layard, R., and Woodhall, N. (1969). The Cases of Gradate Unemployment in India, London, Pengin Books, Bowles, S. (1970). "Migration as Investment: Empirical Tests of the Hman Investment Approach to Geographical Mobility," Review of Economics and Statistics, November, Fields, Gary S. (1972). A Theory of Edcation and Labor Markets in Less Developed Contries, Unpblished Doctoral Dissertation, Department of Economics, University of Michigan, Ann Arbor, Frank, Charlen %., Jr.,1971). "The Problem of Urban Unemployment in Africa" in Ridker, Ronald G. and Lbell, Harold, ed., Employment and Unemployment Problems of the Near East and Soth Asia, London, Vikas Pblications, Cgler, J. (1968). "The Impact of Labor Migration on Society and Economy in Sb-Saharan Africa: Empirical Findings and Theoretical Considerations." African Social Research, 6, December, Harris, John R. and Todaro, Michael P. (1970). "'iration, Unemployment and Development: A Two-Sector Analysis," American Economic Review, March, 1970, pp Johnson, George E. (1971)."The Strctre of Rral-Urban Migration Models," Eastern Africa Economic Review, Jne, 1971, pp Kreger, Anne (1971). "Trkish Edcation and Manpower Development: Some ImpressAons," in Miller,Dncan R., ed., Essays on Labor Force and Employment in Trkey, 1971, pp

29 -27 S~orov, George (1969). "Highlights of the Symposim," UNESCO, Manpower Aspects of Edcational Planning, Paris, Stiglitz, J.E. (1969). "Rral-Urban Migration, Srpls Labor, and the Relationship between Urban and Rral Wages," Pastern Africa Economic Review, December, Todaro, Michael P. (1968). "An Analysis of Indstrialization Employment and Unemployment in Less Developed Contires," Yale Economic Essays, Fall, Todaro, Michael P. (1969). "A Model of Labor Migration and Urban Unemployment in Less Developed Contries," American Economic Review, March, 1969, pp Todaro, Michael P. (1971). "Edcation and Rral-Urban Migration: Theoretical Constrcts and E~pirical Evidence from Kenya," Paper prepared for a Conference on Urban Unemployment in Africa, (nstitte for Development Stdies, University of Sssex, September, frnham, David (1971). The Employment Problem in Less Developed Contries, Paris, Development Centre of the Organisation for Economic Co-operation and Development, 1971.