THE WORLD BANK Internal Discussion Paper SouT ASIA REGION Report No. IDP-121 Labor Entrenchment and Redundancy Compensation in State Owned Enterprises: The Case of Sri Lanka Ariel Fiszbein December 1992 Office of Chief Economist SASVP The views presented here are those of the author, and they should not be mterpreted as reflecting those of the World Bank SOUTH ASIA REGIONAL SERIES Title Author Date Originator IDP 111 How Composition of Publo Expenditure Affects Competitiveness: The Case of Bangladesh K. Matin March 1992 P. Mitra (80419) IDP1 21 Labor Retrenchment and Redundancy Compensation in State Owned Enterprises: The Case of Sri Lanka A. Fiszbeln December 1992 G. Nankani (84641) ASIA REGION DISCUSSION PAPERS 1/ rtile Author Date Oridnator IDP2 The Labor Force Participation of Women in the Republic of Korea: Evolution and Polioy Issues C. Grootaott May 1988 F. lqbal IDP1 5 The Role of Exchange Rate Policy in Four East Asian Countries S-W. Namn May 1988 D. Leipziger (81388) IDP28 The Small-Sole Enterpriso Credit Program (S.S.E.P.) Undor the Second and Third Caloutta Urban Development Projects (CUDP II and CUOP l1l) - An Assessment F. Kahnert March 1988 F. Kahnert (81413) IDP35 Improving Tax Policy Advice: Lessons and Unresolved Issues from Asia Experience H. Feiosig June 1989 H. Flisisig (81413) IDP36 Direot Taxes and Fiscal Polioy Issues: An Illustration for East Asia A. Virmani June 1989 H. Feiesig (81413) IDP37 Commodity Taxation in Selected Countries in South East and East Asia Z. Shalizi June 1989 H. Fielsig (81413) IDP38 Tax Analysis in Developing Country Sottings R. Musgrave June 1989 H. Flelsig (81413) IDP39 Indonesia: Extemal Shocks, Policy Response and Adjustment Performance S. Ahmed June 1989 S. Ahmed (82467) 1/ Papers still available upon request. Asia Sorioes discontinued following December 1, 1991 restructuring of the Region. Note: Extra copies may be obtained from the Asia Information Service Center. LABOR RETRENCHMEM AND REDUNDANCY COMPENSATION IN STATE OVNED ENTERPRISES: THE CASE OF SRI LANKA Ariel Fiszbein SASVP World Bank Revised: December 1992 This paper was prepared as a possible input for the proposed Sri Lanka Economic Restructuring for the Credit. However, the underlying conceptual approach to questions of labor retrenchment and redundancy compensation packages may be of interest to other countries. The author acknowledges the assistance of Shankuntala Gunarate in the preparation of Appendix One. Mats Gustavsson's help was instrumental in estimating redundancy rates. Sriyani Hulugalle helped collecting labor market information. The paper benefitted from comments and suggestions made by Gobind Nankani, Orsalia Kalantzopoulos, Menachem Prywes, and Alejandra Cox Edwards. All remaining mistakes are the author's exclusive responsibility. LABOR RETRENCHMENT AND REDUNDANCY COMPENSATION IN STATE OWNED ENTERPRISES THE CASE OF SRI LANKA: Ariel Fiszbein ABSTRACT Labor issues are an important component of the program of privatization and reform of State Owned Enterprises in Sri Lanka There are currently approximately 120,000 people employed in the public enterprise sector (not including the State Plantations). The levels of labor redundancy appear to be extremely high (between 40 and 50 percent according to our estimates). Thus, privatization and reform will necessarily involve labor retrenchment. This paper addresses the question of labor retrenchment and redundancy compensation in State Owned Enterprises in Sri Lanka, and its main purpose is to provide a conceptual framework to analyze alternative strategies and solutions. The issues involved are complex and have political implications. Under current policies, buyers of privatized firms are forced to maintain existing employment levels. This policy reduces the value of those firms, diminishes levels of productivity, and, in many cases, creates additional distortions as firms require subsidies and trade protection to compensate for the high levels of over-staffing they experience. In addition, the lack of a clear legal framework to determine the payment of redundancy compensation and the past use of several alternative packages generates uncertainty about the cost of a labor retrenchment program. The first part of the paper discusses the retrenchment strategy and estimates the extent of labor redundancy in State Owned Enterprises. It argues that, if the privatization program is to be pursued in a consistent manner, firms should be given the flexibility to determine the size and composition of their work-force. It suggests that the question of determining optimal employment levels be uncoupled from that of terminating employment and compensating redundant workers. The .econd part of the paper proposes the principles which should guide the design of a compensation package for redundant workers. Based on the premise that the retrenchment program wil probably be voluntary, it is suggested that the compensation should be equal to the opportunity cost of leaving the enteprise. Using a simple specification for the opportunity cost, a payment schedule is derived and the financial cost of alternative packages is estimated TABLE OF CONTENTS 1. Labor Retrenchment .................... ............. 1 1.1 Retrenchment before privatization: It is justified? ........................1 1.2 Estimating Labor Redundancy . . ....................................... 4 2. Compensation Packages ...... ................. 7 2.1 The Sri Lankan Experience .................................. . . . 7 2.2 A Conceptual Approach . ...10 2.3 Cost Estimates of Alternative Compensation Packages. ..... 17 Appendix 1. Estimates of Labor Redundancy in selected SOE.... 21 Appendix 2. A Model of Redundancy Compensation: An Application to Sri Lanka. .. 25 Appendix 3. Previously Used Compensation Packages. ... ......... 33 LABOR RETRENCHMENT AND REDUNDANCY COMPENSATION 1. LABOR RETRECHMENT The privatization and public enterprise reform programs in Sri Lanka will involve significant labor retrenchment as most State Owned Enterprises (SOEs) experience a high level of labor redundancy. This section of the paper discusses some basic conceptual issues regarding the process of retrenchment and offers estimates of labor redundancy in SOEs. In the case of those enterprises which, at least in the near future, will not be included in the divestiture program, labor retrenchment should be part of an overhaul of their employment practices. Thus, estimating the extent of redundancy by category and designing a timetable for their elimination should be the first step of a program in which labor practices should be reviewed in order to increase productivity levels. In the case of enterprises which the Government is interested to privatize in the near future, it is necessary to clarify the nature of the retrenchment strategy. 1.1 Retrenchment Before Privatzaion: Is it justified? Under what conditions should the public sector undertake a program of labor retrenchment in SOEs which it is planning to privatize? The question is extremely important as such programs involve significant financial, technical, and political stress. This section discusses some of the methodological issues implied by this problem, and suggests that the answer to this question should be based on sound cost-benefit principles. When there exists labor redundancy, the market value of an SOE i- depressed as the efficient operation of the enterprise will require the elimination of such a redundancy with the ensuing costs (severance payments and others). As long as the potential buyers and the public sector face the same costs and have the same information on the optimal size and composition of the work force, the choice between a "pre-privatization" or a post-- privatization" retrenchment program is irrelevant: the retrenchment cost will be reflected in the selling price."' In practice the public and private sector face different costs and have different sets of information. In the case of Sri Tanka, the public sector is not subject to the complex set of labor laws applying to the private sector.Y As a result, it appears that the costs of labor 1' If the costs of eliminating labor redundancy are sufficiently large, it is possible that the market price of the SOE be negative. v' Nevertheless, certain regulations may apply in the case of partic'ilar enterprises as special provisions have been included in their covenants. -2 - retrenchment are lower for the public sector, in such a way that it can derive a rentl' from the process of eliminating labor redundancy in the enterprises to be privatized. It is important to note that, from the point of view of the buyer, there exists a high level of uncertainty, about the true cost of retrenchment which depends on the complex relationship between workers, unions, the labor justice and the labor relations establishment. As a result, the public sector's cost advantage might be even larger as the market value of the firm would involve a risk-premium.' The Government, however, faces also the political costs associated with its actions. In principle it could be argued that such costs (ie. the loss of votes from those affected by the retrenchment) diminish or eliminate the apparent cost advantage just mentioned. However, it is likely that the Government would suffer those costs even if the elimination of labor redundancy is done by the private sector, as the retrenchment program would be perceived as the natural consequence of the Government's privatization process.Y' From a perspective which emphasizes information, it must be recognized that the optimal size and composition of the labor force depend on the buyer's production plans. Factors such as relative prices, technological alternatives and investment plans, might determine an optimal scale of production quite different to the one prevailing at the time of the privatization. It is reasonable to assume that the potential buyer has more accurate information than the public sector on these matters.i Given this source of uncertainty, the state could be either over or under-estimating the extent of labor redundancy, or misjudging the type of skdlls that are redundant. Once the private sector's informational advantage is considered, the possibility that the public sector faces higher expected costs of retrenchment cannot, a priori, be ruled out. In this case, the state is profiting, at the expense of the potential buyers, of the regulatory system it has developed and from which it is exempted. When true uncertainty rather than probabilistic risk is considered, the outcome might be one in which there are no buyers. It can be argued that paying this political cost would show the government's commitment to the Public Enterprise reform program, and thus is a necessary condition to establish credibility. An interesting asymmetry is found between the treatment given to capital and labor in privatization processes. Traditionally, no calls for retrenching the capital stock of the public enterprises are found as, it is assumed, the private sector has better information and knowledge than the public sector on how to conduct such a process. A similar argument can be used in the case of labor. - 3 - In practice, it is probably impossible to determine whether the net costs of retrenchment are in effect lower for the public sector. However, the previous discussion indicates that the cost-advantage faced by the public sector is lower in the case of industries it which the information problem is more serious. In Sri Lanka, the principal gap in the cost of eliminating redundancies arises from the "Termination of Employment Act" (TEA) which makes the process of dismissing workers from the private sector extremely cumbersome. Under the TEA, employers need to obtain the Labor Commissioner's approval for the dismissal of employees based on non-disciplinary reesons. The Labor Commissioner can deny his approval, and determines the amount of the severance compensation above and beyond the mandatory gratuity payment of one half of a monthly salary for each year of employment. The Labor Commissioner's rulings cannot be appealed, and he does not offer any public discussion of his decisions. Government departments as well as public enterprises are not subject to the TEA. However, even though the state is entitled to follow a program of forced layoffs ("involuntary retirement"), in practice political constraints seem to indicate that the retrenchment will have to be based on a voluntary program. The TEA does not preclude the potential buyer from developing a similar program, in which case the supposed cost-advantage faced by the public sector disappears and becomes negative due to the information-advantage the buyer has. Nevertheless, the design of the optimal voluntary redundancy compensation package is quite different when the threat (irrespective of how remote it is) of forced retrenchment exists (as is the case of the public sector) than when it is practically ruled out (as is the case of the private sector). When that threat is absent, the optimal compensation might be higher as workers have more bargaining power. This is probably the main justification for the public sector undertaling a program of voluntary retirement in the case of an enterprise to be privatized. When the option of using forced lay-offs is evaluated, the state has the possibility of reducing the private sector's retrenchment costs by changing regulations. Those changes can also be seen as an investment: the state would face a cost (probably very difficult to quantify) which would add value to the SOEs to be privatized. The expected return to that investment should thus be compared with the rent the public sector is deriving from the existence of the regulations.7' It is the current policy of the Government to require from buyers of privatized SOEs a commitment of maintaining existing levels of employment. This practice not only reduces the sale price of the SOEs but also unnecessarily reduces labor productivity in the private sector. If the principal objective of the privatization program being a more efficient use of 2' A reform of labor legislation and regulatory practices in Sri Lanka is needed independently of the Public Enterprise reform program, in order to improve labor mobility and productivity, as well as to encourage new investment. - 4 - Sri Lanka's resources, forcing firms to maintain non-optimal levels of employment is a self- defeating strategy. If the privatization program is to be pursued in a consistent manner, firms should be given the flexibility to determine the size and composition of their work-force. The tension between a pre-prnvaiaton and a post-privatization retrenchment strategy can be solved by uncoupling the detemination of optimal employment levels from that of the termination of employment and compensation payment. Under this approach, buyers would have freedom to offer employment to as many workers as they wish. It would then be the Government's responsibility to compensate redundant workers in the same way as in the case of those enterprises that are not ready for divestiture. This strategy has the advantage of giving the private sector the freedom to determine the optimal employment level without increasing the uncertainty cost which would result if the private sector were to asked to compensate redundant workers. 1.2 Estimating Labor Redundancy How should one estimate the extent of labor redundancy in specific enterprises? The main conceptual difficulty is establishing the yardstick by which to judge existing employment levels. in practice, the approach traditionally followed is to compare some measure of labor productivity within the firm with some generally accepted technical standard. For example, in the case of a steel mill, productivity (ie. tons of steel per worker per year) can be compared with that prevailing in similar firms in other countries. It is easy to see the difficulties one faces when trying to apply this methodology, particularly when a large number of firms are considered as in the case of Sri Lanka. Availability of information is cetainly a constraint, but deeper problems exist. The choice of the output and inputs is not obvious. In terms of output firms seldom produce a single good, which implies that in most cases one would have to judge the firm on the basis of several measures. Furthermore, in certain cases the output is not clearly determined (what is the output of a trading board?). In terms of inputs, the composition of the work-force matters and not just the total number of employees. However, even if detailed statistics are available, one would either have to calculate productivity measures for different groups of workers (probably very difficult to do) or face the question of calculating a weighted average. Once a measure of productivity has been agreed upon, it is necessary to obtain similar measures to compare them with. These will probably come from other countries as it is highly unlikely that similar enterprises exist within the same country. However, one has to be extremely careful in choosing measures conesponding to enterprises using comparable technology. - - There is a more basic conceptual problem with the "technical" approach, just discussed. Essentially, this approach ignores the fact that the optimal scale of production and relative factor use after privatization might be different from one currently prevailing. If, for example, the privatization is to be accompanied by deregulation and perhaps even trade liberalization, output under the new conditions (rather than current output) should be used in order to estimate the extent of labor redundancy. This, however, would be an herculean task for just one enterprise and simply impossible for a large set of firms. When labor retrenchment is considered in the case of public enterprises which will not be privatized, social welfare maximization should be the criterion used to estimate labor redundancy. Svejnar and Terrel-' argue that, if GNP maximization is used as the objective, shadow wages rather than actual wages should be used in determining the extent of labor .edundancy. In this case, a worker is redundant when his/her marginal productivity is lower than productivity outside the firm. Under this approach a wiorker in a SOEs who is receiving a wage higher than his marginal product is not redundant if her productivity is higher in the fim than outside it.2' Given the type of information available, this paperfollows a very straighIforward technical approach. The estimates of labor redundancy were obtained using aggregated information on output and employment. They should be taken as rough indicaxtors of the extent of labor redundancy in the enterprises being considered. In that sense, they are just a first approximation toward this complex issue. More sophisticated firm-level analyses will be required in order to determine the extent of labor redundancy in a more accurate fashion. Table 1 presents the estimates of labor redundancy in eight of the major public enterprises. A detailed description of how these values were obtained is included in Appendix 1. A measure of labor productivity was estimated for each enterprise (ie. traffic units per employee in the case of the Railways) and compared with the values corresponding to other developing countries. In most cases, redundancy was calculated using average productivity values for comparable countries. Jan Svejnar and Katherine Terel, Reducing Labor Redundancy in State owned antgaRdze, WPS 792, The World Bank, October 1991. 21 This argument points out to an asymmetry between the criterion for redundancy in private firms which maximize profits, and in SOEs which maximize GNP. - 6 - Table 1 Estimates of Labor Redundancy in Selected SOEs COmLanY Redundancy Ceylon Electricity Board 0.51 Lanka Electricity 0.45 Railways 0.48 Sugar Corporation 0.86 Ceylon Petroleum 0.40 Sri Lanka Cement Co 0.46 Lanka Cement Ltd 0.63 Ceylon Shipping Co 0.43 The average redundancy for these eight ent-.rprises is 53%, while the employment- weighted average is 52%. If the Sugar Company is excluded, both the weighted and un- weighted average equal 48%. On the basis of these estimates, it is possible to calculate the level of over-staffing in the rest of the SOEs. Table 2 shows the estimated number of redundant workers in different sets of enterprises under two alternative assumptions regarding over-staffing. Enterprises in the non-utilities group are classified in three categories according to priority in the divestiture list (category one having the highest priority and category tlree the lowest one). These estimates give an approximate dimension of the redundancy problem, and will be used to calculate the cost of compensation packages for retrenched workers. *-7 - Table 2 Labor Redundancy in SOEsl' Employment 40 Redundany 5an Utilities 33490 13396 16745 Non-Utilities 80715 32286 40358 Category 1 50520 20208 25260 Category 2 23129 9252 11565 Category 3 7066 2826 3533 Total 114205 45682 57103 2. COMPENSATION PACKAGES 2.1 The Sri Lankan Experience Labor laws in Sri Lanka do not establish a clear framework to determine the payment of redundancy compensations. In the case of private sector firms employing 15 or more workers, the Termination of Employment of Workmen Act of 1971 (TEA) establishes that unless an employee provides written consent, employers need to obtain the approval of the Labor Commission for retrenchment of labor. The Labor Commissioner has the authority to grant permission and to determine the amount of the compensation. His decisions are fully discretionary to the extent that no scale or objective criteria exists, and he is not required to explain his rulings which are unappealable. In practice, the Labor Commissioner considers the enterprise's financial means when determining the compensation. The mandatory retirement-gratuity payment of half a month's salary per year of employment, to which all firms are subject to, appears to be the baseline for redundancy compensation payments. iw This list includes 95 SOEs. Plantations are not included. -8 - Public sector employees (including civil servants and employees in state-owned enterprises) are not covered by the TEA. Thus, legally, the state has the right to dismiss workers,"1' while no set criteria exists for the payment of a redundancy compensation. in practice, however, the public sector in Sri Lanka has used voluntary retirement schemes when dealing with labor redundancy. Several compensation packages have been offered to workers in SOEs in the recent past. Faced with the prospect of having to deal with labor redundancy, the Government appointed a committee to study the payment of compensation for redundant staff in Government Corporations and Statutory Boards. The Committee recommended a compensation package popularly known as the Bulumulla Package. The idea behind the BulumuLla package was to have a uniform policy for all redundant public sector employees. Workers would receive half a month's salary per year of employment (the gratuity payment) and a compensation for "denied service" (the period between retenchment and the worker's 55th birthday), which was dermined according to the following formula: one month's salary x (55-Age) x (Age/55) x (Years of Service/30)' The formula had the characteristic that the amount of the compensation decreased with age (maintaining seniority constant) reflecting the adjustment for denial of service. The Bulumulla package was not very popular among its potential recipients. It is unclear whether its failure was due to the fact that it genuinely under-compensated retrenched workers, or because more attractive packages started to be offered immediately after. A second package offered by the Government, which became very popular, is known as the "Leather Corporation" package, an SOE which was privatized. This package was significantly more attractive than the Bulumulla one. Table 3 compares the two packages. IN Workers have the right to file suits against the state as an employer. LW The package included a minimum payment of six month's salaries which, given the formula, everyone with less than seven years of employment would receive. There was also a maximum compensation, but only workers 54 years old with at least 23 years of service would be affected by it. -9 - Table 3 Alternative Comnsatona Years of service Monthly Salaries Bulumulla Leather Cgrporaionl3' 1-4 6 10 5-9 6-9 25 10-14 6-13 35 15-20 8-14 45 20+ 12-19 50 The private sector has also offered compensation packages for redundant workers. One interesting case is the compensation offered by the management of the newly privatized Leather Corporation, which was significantly less attractive than the one offered by the Government before privatization. In fact, under this scheme, the payments were closer to the Bulumulla han to the original Leather Corporation package. This indicates that the original package was probably above the opportunity cost for workers. In other words, if the Bulumulla package "under-compensates" workers, the original Leather Corporation package "over-compensated" them. I' Under current circumstances, it is clear that the Government does not intend to forcibly retrench workers in SOEs. During the month of May of 1992, the President of Sri Lanka announced an employment guarantee for all public sector employees until they attain the age of fiy five. This implies that the reduction or elimination of labor redundancy will have to take place through a process of voluntary retirement. In that context, the design of a compensation package which, while being generous enough to encourage resignations, but at the same time is not unduly expensive, becomes crucial. Given the existence of a variety of packages (and the uncertainty about future ones) there have been mounting expectations on the amounts of compensation which can be obtained The package had a maximum compensation of Rs200,000. This cap probably did not affect many workers. According to the 1990 Census of Public Sector employees, 91% of all SOE employees earned less than Rs 4,000 a month, in which case the cap would not affect them independently of their years of service. The evidence on "market-determined" compensation packages is, nevertheless, mixed. Appendix 3 reports a series of compensation packages offered by firms in the private sector. Those packages vary significantly and it is not simple to compare them. However many of these packages appear to be more generous than the Bulumulla package and similar to the original Leather Corporation package. Thus, the evidence is inconclusive. - 10- from the Government. For any package to be successful in reducing labor redundancies at a reasonable cost, a consistent policy is imperative. This implies that a single package must be offered to all public sector employees in order to avoid a negotiating game which encourages the occurrence of a ratchet effect on redundancy compensation packages. In accordance with the retrenchment strategy discussed in section 1.1, in the case of enterprises in the divestiture list, the alternative of voluntary retirement should be open only after privatization and exclusively to those workers who are not offered employment by the new management. Finally, it is possible that the compensation will have to be more generous than the Bulumulla package, but probably not as generous as the Leather Corporation one. It is necessary to have a more rigorous method to determine the compensation package. To that we turn our attention next. 2.2 A Conceptual Approach 2.2.1 A Review of Alternative Approaches The economic literature on redundancy compensation is not very extensive. Most actual compensation packages are based on existing laws or are the result of specific political and bargaining conditions. The approach found in Svejnar and Terrelli' discussed in section 1.2 looks at the problem from the point of view of GNP maximization. Thus, it does not consider the question of whether the compensation is attractive for workers, which implicitly implies one is dealing with involuntary dismissals. In their approach, a worker is redundant if his marginal product is higher outside than inside the firm. The amount of the compensation is exogenously given. The net benefit to society (expressed in terms of output gains) is calculated as the present discounted value of the productivity gains and the net severance cost. These calculations yield an estimate of the payback period, an indicator of the total cost of the retrenchment program. The literature on labor contracts has shown that optimal contracts (both implicit and explicit) can include provisions for redundancy compensation. When workers are risk averse, a severance pay is part of a contract in which the firm smooths out wages over states of nature.' In some efficiency wage models, wages are found to increase over time as a way to discourage shirking on the part of workers. In that case, the severance payment compensates lSt Svejnar and Terrell, op.cit. 6y Grossman, S. and Hart, O., 'Implicit contracts, moral hazard and unemployment', American Economic Review, vol.71, 1981. - 11- workers for their expected wage loss in case they are involuntarily dismissed.il' Finally, when there exists firm-specific training with uncertain returns, the optimal contract implies that workers will bear part of the training cost. In that case, as in the case of efficiency wages, workers must be compensated if dismissed.)w None of the above approaches considers the question of an ex-post compensation in the case of voluntary labor retrenchment. In the next a simple example of such an approach is discussed. 2.2.2 An Opportunity Cost Approach to Redundancy Compensations This section discusses the main elements to be considered in the design of a redundancy compensation package in the case of voluntary retrenchment. As already indicated, the Sri Lankan political environment is such that there exists an implicit (and in some opportunities explicit) public employment guarantee. Thus, the approach followed here takes existing public jobs as a type of entitlement. In that sense, the redundancy compensation should be seen as a second best solution given the strong social and political constraints under which the retrenchment process will have to take place. Given the voluntary nature of the program, it is possible to disregard legal implications as well as "fairness' considerations. The approach taken implies asking what is the minimum compensation a worker must receive in order to voluntarily leave the enterprise. Thus, the compensation must be equal to the worker's opportunity cost of leaving the enterprise, which in turn is equal to the present discounted value of his expected lifetime income loss resulting from the job loss.'9 It is natural to start this discussion by asking what is the nature of the pecuniaryW loss. Three types of losses should be considered. First, net earnings in alternative jobs might be lower. Here, the relevant concept of earnings includes not just salaries but also fringe benefits and tax advantages. When all these factors are considered, it is an accepted fact that for most individuals there is a positive premium associated with employment in the public sector in Sri Lanka. 17/ Lazear, E., "Agency, earnings profiles, productivity and hours restrictions", American Economic Review, vol. 71, 1981. Boot, A. and Chatterji, M. "Redundancy payments and firm-specific Waining", Economic, vol.56, 1989. In the case of forced retrenchment, the opportunity cost represents the upper bound for the compensation, considering the government's need to minimize financial costs. ;' It could be argued that resigning from a SOE involves, in addition, nonpecuniary costs as "loss of prestige" or others which are not measurable. In what follows those costs will not be considered. - 12 - Second, the process of finding a new job is time-consuming and as a result an initial period of unemployment is likely to result after retrenchment. Given the absence of a system of unemployment compensation, the individual would not receive any pay during such a period. Finally, while public sector employment offers job security, private sector employment involves a certain probability of involuntary separation.A' As a result, the expected alternative wage is lower than actual wages. When all these factors are adequately combined, it is possible to obtain an explicit expression for the opportunity cost which could be used as an indicator for an optimal compensation payment. A mathematical derivation of such an expression is shown in Appendix 2. In essence, the opportunity cost for individuals with different ages is expressed as a function of several factors including wages in the public sector, wage differentials, and the probabilities of finding and losing private sector jobs. The amount of the compensation increases with the public sector wage (both the current and the future) and the public/private wage differential, and diminishes with the probability of finding and retaining private sector jobs. In this scheme, the amount of the compensation is related to the age of the individual in severl ways. First, the older the individual is, the lower the number of years of wage loss for which she will have to be compensated. For example, consistently with the perception of jobs as entitlements, a 54 year old employee would only have to be partially compensated for an additional year of work while a 34 year old one would have to be parally compensated for twenty years. It must be recognized, however, that this compensation excludes pension p.mts which increase with . Thus, while an older worker is compensated for less years of "denied' employment, he will receive a larger pension payment at the time of retirement.Llf Second, the probability of finding a new job and that of retaining it may be associated with age. The three key factors influencing these probabilities are the age-specific rates of unemployment, employment growth and labor turnover. In Sri Lanka, as in probably any other country, both unemployment and employment growth rates diminish with age. Specific i/ I This statement should be qualified for two reasons. First, the guarantee of public sector job security is only formal. That is, in principle there exists the possibility of forced dismissal. In fact, it could be argued that a rational individual should realize that forced retrenchment will occur if not enough workers apply for the voluntary program. Second, current labor regulations in Sri Ianka impose significant restrictions on the private sector in terms of dismissals. Thus, it could be argued, the public/private differential in job security is not very large in the case of Sri Lanka. 21' See section 2.3 for a discussion of pension payments in Sri Lanka. - 13 - values of these two rates for Sri Lanka are reported in Table A2. 1 in Appendix 2. Given those values, the probability of finding a new job in Sri T anka in a one-year period is found to follow an inverted U-curve pattern: it increase until the age of 30, remains constant until age 49, and diminishes after that. Unfortunately, no information is available that would allow even the most basic estimation of an age-profile of labor turnover rates. Third, public/private wage differentials might change with age. However, in pracice, it is very difficult to determine the age-profile of wage differentials as one would have to control for differences in job-tenure, a variable which is seldom measured in econometric studies. Overall, it is not possible to establish a simple qualitative link between age and compensation, as the relationship depends on a variety of factors. Thus, the answer must be quantitatively determined. For that, the model in Appendix 2 was simulated empirically. The data and assumptions used in the simulation as well as the detailed results are shown in the Appendix. Table 4 presents some of the most relevant results. The second column indicates the opportunity cost in terms of annual salaries at time of retrenchment. The third column is a transformation of the second, where the opportunity cost is expressed as annual salaries at time of retrenchment for a 20 year-old worker. Before discussing the results from the simulation, it is !mportant to discuss the role of seniority in affecting the opportunity cost. The approach discussed in this paper is forward- looking in the sense that what must be compensated is future earnings differentials and not past behavior. Thus, seniority (understood as job tenure) will matter only to the extent that it affects the expected future income stream.A' Lets assume there is a return to seniority (independent of age). wages increase with on- the-job experience. If returns are constant (wages increase by x% a year) the opportunity cost is independent of seniority as the expected evolution of wages can be determined exclusively on the basis of the individual's current wage and age. If returns are increasing so is the loss 2y Seniority would also affect the opportWuity cost if the probability of finding a job were to vary with it. Sever arguments can be made in this espect. On the one hand, it could be argued that the "ability to find jobs" diminishes when one does not exercise it. On the other hand, seniority can be perceived as a signal of "stability" and "good work habits". In the first case the probability of finding jobs diminishes with seniority while in the other it increases. Also, if returns to seniority in the public sector are independent of changes in productivity, the more senior workers will have a higher reservion wage which does not necessarily reflect higher productivity. Under those circumstances the expected search time would increase with seniority. - 14 - Table 4 Estimates of Compensation Payment by AMe: Opportunity Cost Method Age Number of Number of Annual Salaries Annual Salaries at Age 20 20 3.55 3.55 21 3.38 3.44 22 3.20 3.32 23 2.99 3.17 24 2.77 2.99 25 2.52 2.77 26 2.44 2.74 27 2.35 2.69 28 2.25 2.62 29 2.14 2.55 30 2.01 2.45 31 1.98 2.46 32 1.95 2.48 33 1.93 2.49 34 1.92 2.49 35 1.89 2.50 36 1.88 2.50 37 1.85 2.50 38 1.83 2.50 39 1.81 2.50 40 1.78 2.49 41 1.76 2.47 42 1.72 2.45 43 1.68 2.43 44 1.63 2.40 45 1.58 2.36 46 1.52 2.30 47 1.46 2.24 48 1.39 2.17 49 1.32 2.08 50 1.25 1.98 51 1.05 1.68 52 0.83 1.34 53 0.58 0.95 54 0.31 0.51 - 15 - that must be compensated, while if there are diminishing returns the loss to be compensated is negatively related to seniority.' In practice, we know very little about returns to seniority. It is very likely that returns to seniority change discretely rather than continuously and are not monotonic. For the purpose of the simulation the case of constant (and continuous) returns was considered, in such a way that one can concentrate on the age-profile disregarding the independent effect of seniority. In the simulations, the opportunity cost, and correspondingly the compensation payment, is found to diminish with age. It diminishes from approximately three and one half annual salaries for a 20 year-old worker (with 35 years to retirement) to less than one third of an annual salary for a 54 year-old worker (with 1 year to retirement). The opportunity cost for the average worker in the public sector (aged 36) is found to be equal to less than two annual salaries. The results do not appear to be sensitive to the value of the probability of retaining jobs, the unemployment rate or the rate of growth in employment.m The opportunity cost, however, is found to be very sensitive to the public/private wage differential.7' In terms of the three factors which determine the existence of an opportunity cost, the simulations show that the existence of a period of unemployment following retrenchment, and the probability of involuntary separation in private sector jobs do not significantly affect the magnitude of the opportunity cost (and the compensation payment), while the magnitude of the opportunity cost is found to be strongly dependent on that of the public/private wage differentialA.Y It is interesting to compare the estimates of the opportunity cost with the compensation payments under the Bulumulla and Leather Corporation packages. As the latter are dependent on years of service, it is necessary to make an assumption regarding the relationship between age and years of service. The average opportunity cost for the 31 to 40 years of age group was compared with the compensation payments under the two packages, assuming three alternative values for the average age at joining service. The results are reported in Table 5. They indicate that, if the opportunity cost Consider the case of two workers of the same age but different seniority. If returns to seniority are increasing, the wage differential between them should rise over time. Thus, the worker with more seniority should receive a proportionately (in terms of current wages) larger compensation. It should be remembered that the cost is expressed in terms of salaries at the time of retirement, which depend on age. So, for example, the opportunity cost, expressed in monetary terms, for a 20 year old is 41 percent higher than for a 36 year old. See Appendix 2 for a more careful description of different simulation exercises. For example, in the case of a worker of age 36, a difference of ten percentage points in the differential, implies a t7 percent change in the opportunity cost. For example, in the case in which there is no wage differential (oa = 1), the compensation for a 36 year-old worker would be only 1.5 monthly salaries. - 16 - approach is used as the basis for calculating the compensation payment, the resulting package is more generous than the Bulumulla but less generous than the original Leather Corporation one. Table 5 mlustQtive Cmnarative Compensation lym=e : Workers in the 3140 Ane Group (Monthly Salaries) Age at Joining Service 20 22 25 Bulumulla Package 14.1 12.3 9.5 Opportunity Cost 22.6 22.6 22.6 Approach Leather Corporation 41.5 38.0 32.0 Package The downward sloping age profile of the estimated opportunity cost is a result which merits some additional comments. Firstly, it should be noted that such a result is not intrinsic to the opportunity cost approach but an empirical finding based on evidence available for Sri Lanka. Secondly, the opportunity cost compensates workers exclusively for 'denied service". In addition, workers would be compensated for past employment through their pensions, in such a way that total compensation will not necessarily diminish with age. Finally, it could be argued that the optimal compensation package should be based on estimates of the social opportunity cost rather than on the individual one as was done in this paper. For example, from a social point of view it might be desirable to pay lower compensations to younger workers in order to lower their reservation wages and decrease their optimal job-search time. However, the perception of public sector jobs as an entitlement of those currendy holding them imposes a limit to an approach based on the social opporunity cost and forces the use of the individual one as the right measure. The opporunity cost can serve as the basis for the design of a redundancy compensation package. However, depending on other objectives, the Government can alter the opportunity cost schedule in order to determine actual payments. For example, a minimum age requirement or a maximum payment cap could be used in order to restrict the scheme to relatively older workers. The Government can affect the opportuity cost itself through changes in its future wage policy. For example, under the retrenchment strategy discussed in section 1 of this paper the Government can reduce the opportunity cost by offering those workers which prefer not to retire a salary fixed in nominal terms until retirement. In any event, the opportmity cost schedule provides the basic knowledge the Government needs to make an informed decision. 2.2.3 Some Additional Comments The previous discussion implicidy assumed workers are homogeneous. In fict, workers are heterogeneous in terms of their skill level. Tbis is due both to ability and training not captured by the - 17 - information on their position within the firm. Heterogeneity has important implications for the process of retrenchment, as an homogeneous compensation package;' is more attractive for the most productive workers who could probably find better paid jobs faster. In the case of the firm to be privatized, the retrenchment strategy discussed in section 1.1 solves this incentive problem. The choice of whether to accept the package is conditional on the worker not being selected by the new management. In other words, only those workers that do not receive a job offer in the privatized firm qualify for the compensation package. Once the enterprise is sold, the quality of the remaining employees is irrelevant to the extent that the state has no incentives in maintaining their employment. The incentive problem is more serious in the case of those firms which are not going to be privatized. In those cases, it is in the firm's interest to offer the retirement option selectively to the "bad" workers.2' Also, the announcement of a new wage policy which compensates on the basis of performance can deter "good' workers from leaving.3" A final point to be discussed is the possibility of making compensation dependent on employment status. This would involve the payment of an initial sum and a monthly payment during the period in which the worker is unemployed. However, in order for such a scheme giving incentives for job search, there must be a time limit for compensation which should decrease over time. The scheme discussed above solves this problem in a simpler way. By calculating the compensation on the basis of the length of unemployment for the average worker it partially covers the losses associated with job search while at the same time provides incentives for efficient search. 2.3 Cost Estimates of Alternative Compensation Packages In this section, rough estimates will be made of the total cost of using three alternative packages (the Bulumulla, the Opportunity Cost, and the Leather Corporation) to compensate retrenched workers. 22' An heterogeneous package would be one in which the amount of the compensation varies with workers's quality. In practice, such a scheme is unfeasible because of imperfect information and the lack of objective performance cnteria on the basis of which the compensation could be determined. In this case, the compensation should be calculated not in terms of average altemative wages and probabilities but in terms of those wages and probabilities that would apply to "bad" workers. However, in order for it to work, this policy should involve higher costs for "bad" workers. Thus, for example, the new and more attractive policy should be such that wage promotion be based on merit/performance. - 18- The estimates will be based on two hypotheses regarding the extent of redundancy (40 and 50 percent). The same classification of SOEs as in Table 2 will be used. The cost will be estimated considering an average employee: a 36 year old workers, with 11 years of service, eaming Rs 3,500 a month.W The results are reported in Table 6. As expected, the opportunity cost package is approximately twice as expensive as the Bulumulla Package, and two thirds as expensive as the Leather Corporation one. The total cost is very significant. Under the 50 percent redundancy assumption, the opportunity cost package would cost a total of approximately US$100 million or the equivalent of 1.6 percent of GDP. The categorization of SOEs shown in Table 6 gives a rough idea of how that cost would be distributed over time (category 1 being those enterprises high in the priority divestiture list). An additional factor which must be considered, is the potential cost the program of labor retrenchment could have on pension funds. SOEs must make contributions to the Employees' Provident Fund (EPF) and the Employees' Trust Fund (ETF). Contributions to the EPF amount to 12 percent of the wage bill, while those to the ETF to 2 percent. In addition, employees contribute 8 percent of their salaries to the EPP. It is uncertain the extent to which SOEs have complied with their obligations. In the case of the EPF, benefit claims (a lump-sum payment) can be made, with few exceptions, at time of retirement (55 years of age). Thus, labor retrenchment in SOEs should not represent any additional unplanned burden on EPF. Also, retrenched workers would not lose their rights to claim benefits as the contributions made by their new employers would simply be added to their past contributions. In the case of the ETF, employees can claim benefits (a lump-sum payment) on termination of current employment. This implies that the ETF would face an unexpected increase in the demand for benefits. Thus, a more careful look at the financial situation of the ETF and the implications of massive retrenchment on its functioning will be required. ;' 7The values on age and seniority were obtained, as averages, from the Census of Public Sector Employees. The average salary was calculated on the basis of information provided by a set of SOEs, and is consistent with information obtained from the Cental Bank. - 19 - Table 6 Estimated Costs of Alternative Compensation Packages (Million Rs.) 40% Redundancy Bulumulla Opport, Coa Leathr5QgM Utilities 472 1058 1641 Non-Utilities 1137 2549 3955 Category 1 712 1596 2475 Category 2 326 731 1133 Category 3 100 223 346 Total Million Rs. 1608 3607 5596 Million US$ 37.8 84.9 131.7 % GDP 0.6 1.2 1.9 50% Redundancy Bulumulla OpDo. Cost Leater Corp. Utilities 590 1322 2051 Non-Utilities 1421 3187 4944 Category 1 889 1995 3094 Category 2 407 913 1417 Category 3 124 279 433 Total Million Rs. 2011 4509 6995 Million US$ 47.3 106.1 164.6 % GDP 0.7 1.6 2.4 For a 36 year old worker, with 11 years of service, earning Rs. 3,500 a month. 1US$ = 42.5 Rs 1991 GDP = US$ 6,785 Million - 20 - APPENDICES - 21 - A1.1 Appendix 1 Estimates of Labor Redundancy in Selected SOE In order to estimate the extent of labor redundancy, indicators of labor productivity were calculated for eight state-owned enterprises. Those indicators were compared with similar ones for other developing countries. In the case of the Railways the indicator used was millions of traffic units (the sum of passenger and goods kilometrage) per employee ratio. On the basis of information provided by the Railways Department, such ratio was found to be 0.14 in 1990. This value compares very poorly with similar indicators found in other developing countries. Table Al .1 shows information for a set of 20 developing countries. It can Table All: Railways: Traffic Units per Employee Country TU/Employee Morocco 0.53 Romania 0.46 China 0.39 Poland 0.39 Malaysia 0.38 Algeria 0.32 Turkey 0.29 Tunisia 0.29 Cote D'Ivoire 0.27 Yugoslavia 0.25 Hungary 0.23 Cameroon 0.21 Pakistan 0.21 India 0.18 Albania 0.16 Sri Lanka 0.14 Kenya 0.13 Bulgaria 0.12 Bangladesh 0.10 Tanzania 0.08 Average 20 countries 0.26 Source: The World Bank Railway Data Base (RDB) - 22 - A1.2 be seen that Sri Lanka is among the countries with the lowest productivity. If the average value is used as an indicator, Sri Lankan Railways show a redundancy of 46%.3' The Ceylon Electricity Board (CEB) is responsible for electricity generation and transmission, as well as for part of the distribution to final consumers. In 1990 it employed 11.1 employees per MW of installed generating capacity. By comparison, the average for a sample of 35 developing countries shows only 5.4 employees per unit (see Table A1.2), which implies a redundancy of 51%. In the case of Lanka Electricity Company (Private) Ltd. (LECO), a distributor, the customer/employee ratio is a good indicator of performance. Currently, LECO has 156 customers per employee. As a reference, India has a ratio of 250 customers per employee and ratios of 300 are not unusual among developing countries. The Sri Lanka Cement Corporation (SLCC) produced in 1991 189 million tones of cement per employee. Other plants in Asia have productivity indices that range between 350 and 1460 MT per annum per employee. The world average is still higher: between 1200 and 1500. Thus, if compared with the lower end of the Asian average, SLCC presents a redundancy of 46%. During the 1988-1990 period, Lanka Cement Ltd (LCL) recorded an average labor productivity figure of 130 MT per annum per person. The highest ever labor productivity at LCL was achieved in 1986, at 180 mt per person. LCL did not operate at all in 1991. Table A1.3 presents the estimates of labor redundancy in three SOE in addition to the five already discussed. This estimates are based on information collected during interviews with managers in those companies. The average redundancy for these eight enterprises is 53%, while the weighted average is 52%. If the Sugar Company is excluded, both the weighted and unweighted average equal 48%. 331 Tbe same indicator for OECD countries ranged between 0.28 and 4.3. - 23 - A1.3 Table Al.2: Labor Productivity in Electricity Generation Country Employee per unit capacity Paraguay 0.39 Venezuela 1.01 Korea 1.15 Zaire 1.40 Mozambique 1.60 Zambia 2.13 Zimbabwe 2.14 Mexico 2.80 Yugoslavia 2.81 Colombia 3.28 Indonesia 3.38 Peru 3.67 Morocco 3-77 Brazil 3.95 Cote D'Ivoire 4.07 Dominican Republic 4.15 Cameroon 4.46 Malaysia 4.49 Nigeria 4.53 Jordan 4.85 Philippines 5.56 Ghana 5.58 Panama 5.61 Hungary 5.68 Tunisia 5.70 Honduras 5.84 Costa Rica 6.26 Syria 6.32 Uruguay 7.65 Ecuador 7.70 Thailand 8.73 Bangladesh 9.26 Sri Lanka 11.14 Kenya 11.65 China 15.58 Pakistan 21.82 Average 5.4 Source: Lee Catalano (Ed.), Directory of Electric Utilities in Developing Siona, Mc Graw Hill, 1986, ard Power Data Sheets, The World Bank. A1A Table Al.3: Labor Redundancy in Selected SOE Company Employment Output Measure Out Productivity Redundancy Sri Lanka Comparable (%) Employees CEB (*) 14180 Capacity (MW) 1280 11.08 5.4 0.51 7268 LECO (*) 1166 Customers 191729 164.00 300.0 0.45 527 Railways (**) 21678 TU (mill.) 2944.7 0.14 0.26 0.48 10352 Sugar Corp. (***) 5306 - (A) - 0.86 4550 Ceylon Petroleum 6107 (A) - 0.40 2443 SLCC (*) 3280 Cement (MI) 619652 189.00 350 0.46 1510 LCL (***) 922 Cement (MI) 119860 130.00 350 0.63 580 Ceylon Shipping 433 - (A) 0.43 185 a Co (** (*) 1991, firm's information. (**) 1990, firm's information. 1990, Census of Public Sector Employees. (A) Estimates based on interviews. - 25 - A2.1 Appendix 2 A Model of Redunan Compensan: An Aplication to Sri Lanka In this appendix a model is built to determine the optimal redundancy compensation. In a voluntary retrenucment progam, such payment shold be enough to make it desirable for the wrker to resign from his job. lTus, it must be equW to the worker's opportunty cost of leaving the terprise. In the case of forced retrenchment, the opportnity cost represens the upper bound for the compensation when the government wishes to minimize financial costs. The existence of an opportunity cost is based on three factors. First, net earnings might be lower in altenative jobs in the private seor. ' The correct measure of emnings is that which includes wages, finge benefits, and differential taxing. Second, an initial period of unemployment might occur following Dung tis period Ihe wovk receves no pay (we wM asme no uao exists, as is the case in Sri Lanka). Third, while jobs in the public sector offer job security, there is a non- negative probability of involuntary separation in alternative jobs. When all factors are taken into account, the opportunity cost equals the present discounted value of the expected wage loss. For a worker "a" years old, the opportunity cost (C8) is given by: C a t -Rat4a)(l 4 h - (1+ (1) where W.(t), R.(t) and Pa(t) are, respectively, the wage in the public sector, the alternative wage and the probability of being employed of a worker of age "a' in period t. T is the officially determined retirement age, and d is the discount rate. We will express the existence of a positive wage differential between the public and private sector in the following manner: R,(t)= Wd(t a (2) The differential could itself vary with age. However, for simplicity, considering the unavailability of data needed for application to the Sri Lankan case, a will be taken as an age-invariant parameter. i! This is not the case in all developing countries. However, although there is little statistical evidence, it is generally agreed that in Sri Lanka wages for most workers (particularly for unskilled workers) are higher in the public than in the private sector. - 26 - A2.2 The probability of an individual of age "a" being employed In a specific point in time can be expressed as the sum of three terms. First, the probability of having been employed the previous period and retaining the job. Second, having been employed the previous period, losing the job, and finding a new one. Third, not being employed the previous period and finding a job in the current period. P441(t =PO1t-I(t-1)r+Pd4t-(t-1)(1 P-r)pt+(l -P4t-1)) p,,Pa 3 where r is the probability of an individual retaining his job and Pa is the probability of an individual of age "a" finding a job in period t. It should be noticed that this last probability varies with age but is not time-dependent. Pa(O) -Pa (4) Through some algebraic manipulation, and considering that the probability of being employed in the initial period is equal to the probability of finding a job in that period (equation (4)), we have: 44t a4t Pa (t pi 1| (r(l -Paf,)) (3') *=a fssM Replacing (2) and (3') into (1) we obtain the following expression: s at aft 4 T-a i I"Ha (r(l-P..))J (1') t-0 (1+df Before discussing the data to be used in the simulations, it is necessary to adopt a definition for the age- specific probability of finding a job (p). We will define the probability of finding a job as the ratio between vacancies and job seekers. In order to obtain an age-specific probability we need to assume the existence of age-specific vacancies. Thus we have: I-r+u4 PO (S) where u, is the unemployment rate among individuals of age "a" (defined in terms of employment rather than in terms of labor force) and gL is the rate of job creation for individuals of age "a". In the simulations, u. and g, will be considered to be constant through time. In the case of the unemployment rate, this implies that it is not affected by the retrenchment itself. To the extent that the rate of job creation increases (decreases) through time, the model will overestimate (underestimate) the opportunity cost. - 27 - A2.3 In order to make this model instrumental it is necessary to define the age profile for three variables: W, g, and u. The model will be used to simulate optimal compensations for archetypical workers in SOEs in Sri Lanka. Thus, rather than assuming specific functional forms for those variables, we will run simulations using alternative assumptions based on, to the extent it is possible, evidence for Sri Lanka. The age-specific unemployment rates were obtained from published data corresponding to the 1991 Labor Force and Socio-Economic Survey. The rate of employment creation (g) was taken as that observed between 1990 and 1991. The wage age-profile was calculated using average3O values from the 1990 Census of Public Sector Employees. In all cases, the values were obtained for five-year age brackets. The values for u., g., and W. are presented in Table A2.1. There are three parameters in the model: r, d and a. The values used in the base case were r=0.04, d=0. 12, and a=0.8. The model was simulated using several other values in order to determine the robustness of the results. Table A2.2 reports the results of the simulation exercises, expresses as yearly salaries at time of retrenchment. The opportunity cost, and correspondingly the compensation payment, is found to diminish with age. It diminishes from approximately three and one half annual salaries for a 20 year-old worker (with 35 years to retirement) to less than one third of an annual salary for a 54 year-old worker (with 1 year to retirement. The opportunity cost for the average worker in the public sector (aged 36) is found to be equal to less than two annual salaries.W The results do not appear to be sensitive to the value of the probability of retaining jobs (r), as indicated by the third column in Table A2.2. Neither the unemployment rate nor the rate of growth in employment show a signiflcant effect on the compensation. A ten percent change in both rates was considered, and the results of the simulation are reported in columns 6 to 9 in Table A2.2. The average elasticity of the opportunity cost with respect to both rates is found to be less than 0.1. That is, a ten percent increase in the unemployment rate increases the opportunity cost by less than one percent. Figures 1 and 2 show the age profile of the opportunity cost under different values of u and g. No significant difference can be appreciated between the different profiles. The opportunity cost, however, is founl to be very sensitive to the public/private wage differential (at). For example, for a worker of age 36, a difference of ten percentage points in the differential, implies a 3./ The micro-data tape was not available at the time this paper was written and thus it was not possible to estimate the profiles econometrically. 3I/ It should be remembered that the cost is expressed in terms of salaries at the time of retirement, which depend on age. So, for example, the opportmity cost, expressed in monetary terms, for a 20 year old is 41 percent higher than for a 36 year old. - 28 - A2.4 Table A2. 1 Values Used in Simulations AGE ua L W* 20 0.57 0.04 1.000 21 0.57 0.04 1.019 22 0.57 0.04 1.039 23 0.57 0.04 1.058 24 0.57 0.04 1.079 25 0.19 0.04 1.099 26 0.19 0.04 1.120 27 0.19 0.04 1.142 28 0.19 0.04 1.166 29 0.19 0.04 1.191 30 0.09 0.09 1.217 31 0.09 0.09 1.243 32 0.09 0.09 1.270 33 0.09 0.09 1.286 34 0.09 0.09 1.302 35 0.07 0.09 1.318 36 0.07 0.09 1.335 37 0.07 0.09 1.351 38 0.07 0.09 1.365 39 0.07 0.09 1.380 40 0.04 0.09 1.394 41 0.04 0.09 1.409 42 0.04 0.09 1.424 43 0.04 0.09 1.446 44 0.04 0.09 1.469 45 0.04 0.07 1.492 46 0.04 0.07 1.515 47 0.04 0.07 1.539 48 0.04 0.07 1.554 49 0.04 0.07 1.570 50 0.04 -0.09 1.586 51 0.04 -0.09 1.602 52 0.04 -0.09 1.618 53 0.04 -0.09 1.634 54 0.04 -0.09 1.650 - 29 - A2.5 Table A2.2 Opportunity Cost (Number of Annual Salaries at Time of Retrenchment) AGE Base r=O.l a=0.9 a=0.7 Ou+ 10% bu-10% bg+ 10% 6g-10% Case 20 3.55 3.56 2.67 4.43 3.66 3.43 3.53 3.57 21 3.38 3.39 2.48 4.27 3.49 3.27 3.36 3.41 22 3.20 3.21 2.28 4.11 3.29 3.10 3.18 3.22 23 2.99 3.01 2.06 3.92 3.08 2.91 2.97 3.02 24 2.77 2.78 1.82 3.72 2.85 2.70 2.75 2.80 25 2.52 2.53 1.55 3.49 2.59 2.47 2.51 2.55 26 2.44 2.45 1.47 3.41 2.50 2.40 2.43 2.47 27 2.35 2.36 1.38 3.32 2.41 2.31 2.34 2.39 28 2.25 2.26 1.28 3.22 2.30 2.22 2.23 2.28 29 2.14 2.14 1.17 3.10 2.18 2.12 2.12 2.17 30 2.01 2.01 1.05 2.97 2.04 2.00 2.00 2.04 31 1.98 1.98 1.03 2.93 2.01 1.97 1.97 2.01 32 1.95 1.95 1.02 2.88 1.97 1.94 1.94 1.97 33 1.93 1.93 1.02 2.85 1.95 1.93 1.92 1.95 34 1.92 1.92 1.01 2.82 1.92 1.91 1.91 1.93 35 1.89 1.90 1.01 2.78 1.90 1.89 1.89 1.90 36 1.88 1.88 1.00 2.75 1.88 1.87 1.87 1.88 37 1.85 1.85 1.00 2.71 1.86 1.85 1.85 1.86 38 1.83 1.83 1.00 2.67 1.84 1.83 1.82 1.84 39 1.81 1.81 0.99 2.63 1.81 1.81 1.80 1.82 40 1.78 1.78 0.99 2.58 1.79 1.78 1.77 1.80 41 1.76 1.76 0.98 2.53 1.76 1.75 1.74 1.77 42 1.72 1.72 0.98 2.47 1.73 1.72 1.71 1.74 43 1.68 1.68 0.97 2.39 1.69 1.67 1.67 1.70 44 1.63 1.63 0.96 2.30 1.64 1.63 1.62 1.65 45 1.58 1.58 0.95 2.21 1.59 1.57 1.56 1.60 46 1.52 1.52 0.94 2.11 1.53 1.51 1.50 1.54 47 1.46 1.46 0.92 1.99 1.47 1.45 1.44 1.48 48 1.39 1.39 0.91 1.87 1.40 1.38 1.37 1.42 49 1.32 1.32 0.90 1.75 1.33 1.31 1.30 1.35 50 1.25 1.26 0.89 1.61 1.26 1.24 1.22 1.28 51 1.05 1.06 0.75 1.35 1.06 1.04 1.03 1.08 52 0.83 0.83 0.59 1.06 0.83 0.82 0.81 0.85 53 0.58 0.59 0.42 0.75 0.59 0.57 0.57 0.60 54 0.31 0.31 0.22 0.39 0.31 0.30 0.30 0.31 - 30 - A2.6 Figure 1 Opportunity Cost AlternatIve UnempIoyment Rates 4 3. * 2.5 - I 1.5 A 0.5 0 I II - I II 20 30 40 s0 AGE + .10% * -10% - 31 - A2.7 Flgure 2 Opportunity Cost Alternative Eploynent Growth Ratee 4 3.5 gs~~ - 2.5- ol2- I~ ~~~ O I~~~~~~~~~~~~~~~~~~~~~~~~ 0.5 20 s0 40 50 10E ;+ +10X o -10X -I . . - 32 - A2.8 47 percent change in the opportunity cost. This effect, however, is stronger in the middle of the age range."' Figure 3 shows the different age profiles derived assuming three wage differentials. In terms of the three factors which determine the existence of an opportunity cost, the simulations show that the existence of a period of unemployment following retrenchment, and the probability of involuntary separation in private sector jobs do not significantly affect the magnitude of the cost (and the compensation payment). The magnitude of the opportunity cost is found to be strongly dependent on that of the public/private wage differential.w Figure 3 Opportunity Cost Aiternatfve waoe Dftforentlale 4.5 4 3.5 2.5 2 1.5 0.5 20 30 40 So AGE + Alpha . 0.9 o Atphs 0.,7 37/ The difference is approximately 25 percent for workers in the two extremes of the age distribution. lit For example, in the case in which there is no wage differential (a= 1), the compensation for a 36 year-old worker would be only 1.5 monthly salaries. - 33 - A3.1 Appendix 3 Previously Used Compensation Packages United Motor Ltd 0 - 1 year Rs2O,000 1 - 3 years Rs3O,000 3 - 10 years Rs60,000 BCC Ltd 0 - 4 12 months 5 - 10 25 months 11 - 15 35 months 16 - 20 45 months 20+ 50 months Ca-lon Oxygn Ltd Rs85,000 + one month's salary per year of service Nestle Lanka Ltd 5 months' salary per year of service Union Carbide Rs3O,000 + one month's salary per year of service - 34 - A3.2 Brook Bod Ld 5 years to retirement 36 months pay + normal gratuity pay 4 years to retirement 23 months pay + normal gratuity pay 3 years to retirement 22 months pay + normal gratuity pay 2 years to retirement 15 months pay + normal gratuity pay 1 year to retire 8 months pay + normal gratuity pay If more than five years to retirement: 1-4 years of service 8 months 5-10 years of service 9-19 months 11-20 years of service 20-36 months 21-30 years of service 38-54.5 months ASIA REGION DISCUSSION PAPERS 1/ (Continued) Title Author Date Oriainator IDP42 An Analysis of the Nature of Unemploymnent W.T. Dickens in Sri Lanka and K. Lang July 1989 R. Zagha (80433) IDP44 Assisting Poor Rural Areas Through Groundwater Irrigation F. Kahnert August 1989 C. Chamberlin (81409) IDP51 Educational Development In Asia: A Comparative Study Focussing on Cost J-P. Tan and and Financing Issues A. Mingat October 1989 J-P. Tan (81408) IDP52 Chinese Reforms, Inflation and the Allocation of Investment in a Sooialist Economy 0. YVnal October 1989 0. Yenal (81415) IDP63 Public Polioy to Promote Industrializa- tion: The Experience of the East Asian NICs and Lessons for Thailand D. Dollar May 1990 D. Dollar (80518) IDP65 A Study of the Poor in Sri Lanka C. Rouse June 1990 Y. Huang (80434) 1DP68 Health Sector Financing In Asia C. Griffin August 1990 J-P. Tan (81408) 1DP74 A Case Study of a Gradual Approach to Economic Reform: The Viet Nam Experience of 1985-88 Z. Drabek September 1990 Z. Drabek (80504) IDP85 On Estimating Inadequacy of Energy Intakes: Revealed Food Consumption Behavior versus Nutritional Norms B.8. Minhas September 1990 S. Jayanthi (81419) IDP88 Asia Region Seminar on Policy Challenges in India October 1990 C. Chamberiin (81409) IDP93 Parametric Population Projection and Its Usefulness for Policy Analysis W. Sanderson April 1991 J-P. Tan (81408) IDP96 A Review of Korea's Trade Pattem D.M. Leipziger and S.Y. Song March 1991 D. Leipziger (81388) IDP97 Sustainability as Intergenerational Equity: The Challenge to Economic Thought and Practice R.B. Norgaard June 1991 F. Johansen (81410) IDP99 Housing Prices, AffordabiUlty, and Government Policy in Korea K-H. Kim July 1991 D. Leipziger (81388) IDP11 Lessons from Experience with Wage Flexibility In Asia I. NabI August 1991 I. Nabi (80514) j Papers still available upon request. Asia Series disoontinued following December 1, 1991 restructuring of the Region. Note: Extra copies may be obtained from the Asia Information Service Center.