World Bank 47305 Employment Policy Primer December 2008 No. 11 UNEMPLOYMENT INSURANCE SIMULATION MODEL (UISIM) Jean Fares and Milan Vodopivec Unemployment is a pressing problem in many China. The model is user-friendly, can be adjusted to parts of the world. It has become increasingly recog- fit the UI system of almost any country, and is power- nized that the process of growth in market economies ful enough to provide accurate estimates for numerous requires intense labor reallocations, with 20 to 30 per- variables of interest. Deliberately preserving similari- cent of jobs being created or destroyed every year. Thus ties to PROST, the World Bank's pension simulation even in well-functioning economies the displacement model,2 UISIM brings numerous advantages over the of workers and ensuing unemployment can be a severe plain spreadsheet models built to simulate an unem- problem. Unemployment problems are exacerbated ployment insurance system. by macroeconomic crises and increased globalization, This note describes the key features of the UISIM which seems to expose more workers to the risk of job model--how the model is structured, what data inputs loss. are needed, and what outputs the model generates (the It is therefore not surprising that the task of helping model comes with user's and technical manuals, which the unemployed has gained importance and attracted provide detailed information about how to operate the the interest of policymakers. Because job loss entails the model and how it calculates the outputs).3 For illustra- loss of income, providing effective income support-- tive purposes, the note also presents an example where while ensuring appropriate work incentives and facili- the model is used to generate simulations for a country- tating access to jobs--is a prime concern and a necessary wide UI system. The appendix to the note describes component of assistance to the unemployed. typical data sources and provides a detailed description UISIM is a model that the Human Development of requisite data. Network at the World Bank expressly developed to assist client countries in making projections of unem- ployment insurance (UI) benefit costs, contributions, 1 Unemployment assistance programs can also be handled by the and trust fund balances.1 The model assesses the program. 2PROST (Pension Reform Option Simulation Toolkit) has gained world- financial status of existing UI schemes under different wide currency and is one of the foremost models of pension reform economic scenarios and conducts financial projections options in developing countries. It too was developed by the Human of newly developed UI programs under alternative set- Development Network of World Bank. 3 tings. Unemployment Insurance Simulation Model (UISIM): User's Manual, Human Development Network, World Bank, December 2004; Unem- The current version of UISIM has been piloted in ployment Insurance Simulation Model (UISIM): Technical Manual, three countries, Bosnia and Herzegovina, Turkey and Human Development Network, World Bank, August 2004. * Note prepared by Katherine Terrell (Ross School of Business and Ford School of Public Policy, University of Michigan) and Rita K. Almeida (Human Development Network Social Protection, World Bank). We thank the comments provided by Jean Fares (HDNSP, World Bank) and Milan Vodopivec (HDNSP, World Bank). Unemployment Insurance Simulation Model (UISIM) Description of the Model unemployment rate is anticipated, the model does not adjust the probability of exiting unemploy- The UISIM model asks the user to complete the ment). base period input data sheets in an Excel workbook, and The model inputs and outputs can be disaggregated then performs calculations to produce output tables by age and gender. and charts. The model projects the primary variables of The basic period of observation is one month, and interest--the UI system's revenues, expenditures, and the time horizon is unlimited. trust fund balances. Written in Visual Basic and C++ compiler and with the interface in Microsoft Excel, the Data requirements model can be run on most new computers and requires UISIM uses the following two types of data (Fig- 10 to 30 seconds to compute the results. Below we ure 1): describe the objectives and key features of the model, as well as the data required for and the outputs generated Economic data by the model. Macroeconomic parameters Population and/or labor market data Objective and key features Data about the UI system Objective. The objective of UISIM is to calculate-- Model settings: modeling and simulation choices simulate--expenditures and revenues of the unemploy- (such as assumptions about the probability of exit- ment insurance system. The model supports the follow- ing unemployment and the simulation horizon). ing key types of simulations: The input of economic data is flexible and allows Introduction of the UI system, allowing compari- entering as many details as the user wants to pro- son of alternative scenarios under different design vide, or, alternatively, very few key data items (such as parameters. total employment and unemployment, average wage) to Modifications of the parameters of the UI system, enable "quick and dirty" simulations (thus relying upon such as changes in initial UI eligibility, length of default values provided by the model itself). potential eligibility, benefit level--including mini- mum and maximum benefits, and indexation of UI Methodology and Output benefits. Changes in financing related to coverage of the UI Once the input requirements are satisfied, the model system and compliance with payments of contribu- follows a simple algorithm to produce the output. Below tions. Impact of labor market changes created by macro- economic shocks and cycles on the performance of Figure 1: UISIM Input Structure the UI system. Economic data Model settings Key features. To be able to simulate the financial 1. Macroeconomic 2. Population Countryname performance of the UI system while retaining a simple parameters Simulationrunname model structure and ease of data input, the following Authorname 3. Labor Market principles and features were adhered to: Startingyear LaborForce Simulationhorizon Unemploymentsurvival Unemploymentdurationlimit Workingexperience UISIM is an accounting model: UI benefits are Localcurrencycode Wages Simulationmethod computed as the number of recipients times the (stock and ow) 4. Unemployment Insurance level of benefits, and similarly, UI revenues are Choiceofinputmodefor Recipients certain variables computed as the number of contributors times Bene ts Contributors the amount of contributions paid. The model Contributions thus does not attempt to model the behavior of Otherlabormarketprograms UIfund individuals (for example, if a larger economy-wide 2 Unemployment Insurance Simulation Model (UISIM) we outline the algorithm and list the output generated mum benefits, which are--as is the case of flat by UISIM. benefits--inputted either in local currency or as a share of economy-wide average or refer- Algorithm ence wage. The user needs to specify time and duration frames for the model. The simulation time frame is Step 6: Calculate survival functions defined by the base-year (e.g., 2008) and the simula- The duration of receipt of UI benefits is a tion horizon y0 (e.g., five years, in this case the simu- crucial determinant of UI expenditures. The lation end-year would be 2012). The basic time unit UISIM model assumes that the survival in of UISIM is a month and the simulation runs from (covered) unemployment--and, equivalently, January of the base-year until December of the end- the hazard of leaving it--is governed by a sur- year. The outline of the algorithm used in UISIM is vival/hazard function. The functional forms of presented below. the duration models used are either a simple exponential function or a Weibull function. Step 1: Calculate smoothly varying monthly values of The user can choose either to model both macroeconomic parameters recipients and non-recipients with one of these survival functions or to model non-recipients Step 2: Project population to each year of the simula- this way. In the latter case, the survival function tion horizon of the recipients is specified through behavioral parameters (by modifying the hazard of exit of Step 3: Determine the initial conditions (i.e., the state the non-recipients). in January of the base-year) Step 7: Calculate new distribution of unemployed Loop the following steps for each month of the To calculate unemployment, first the number simulation: of unemployed in each of the gender and age groups is determined; the survival functions are Step 4: Calculate wages then used to distribute the recipients and non- Wages are calculated based on an inputted earn- recipients according to the duration of their ings profile, which specifies the ratio between stay in unemployment. At any specified date, the average wage of the specific cohort and the user can switch to a flow method where the average wage of a 20 year old male. Over unemployment is regulated by the inflow into the simulation period, wage growth is updated unemployment and survival functions. using the assumed inflation rate and the real wage growth rate. The economy-wide average Step 8: Update labor market tables wage is calculated by aggregating cohort wages weighted by their employment share. Step 9: Calculate finances of the UI fund Step 5: Calculate contributions and benefits Outputs Benefits and contributions are calculated in UISIM generates a series of output tables and charts, a similar manner. For example, benefits can providing monthly or yearly data for the simulation be either flat or variable, or a combination of period. Key variables appearing in the output tables and these. The flat benefits can be entered either charts are: in local currency, as a share of economy-wide average wage or as a share of the reference Number of recipients wage. Variable benefits are inputted as a per- Total revenues (absolute value, share of GDP) centage of the wage that the recipient earned Total benefits (absolute value, share of GDP) in the last month of employment. Variable Contribution rate required to match current rev- benefits are limited by minimum and maxi- enues and benefits 3 Unemployment Insurance Simulation Model (UISIM) UI fund reserves (number of months that the cur- is modified while the others retain their baseline val- rent level of benefits could be supported from the ues (the exception is the "combined reform" scenario, reserves alone). where--as the name suggests--several UI parameters are allowed to change simultaneously). The following Putting UISIM to Use: An Example of UI scenarios are distinguished: System Simulations Increase of benefits scenario, where the level of For illustrative purposes, plausible UI simulation benefits increases from the current A to B percent scenarios for Country X are presented below, and simu- replacement level (a change of flat rates also can be lation results discussed. Of particular interest are setting modeled). An increase of the replacement level from the simulation scenarios, by contrasting various policy 50 to 70 percent is assumed below. scenarios with the baseline scenario, and presenting the Increase of duration scenario, where the potential model's outputs. duration of benefits is increased by XX percent, from the current maximum potential duration of Simulation scenarios YY to ZZ months. An increase of the potential dura- Note that it is assumed that the UI program of tion from 10 to 20 months is assumed below. Country X has been running a surplus and has accu- Easing of eligibility scenario, where the inflow into mulated large reserves. All scenarios below assume 2004 the receipt of UI benefits is doubled and remains at to be historical data input (usually the last year when that level indefinitely. historic data are available), and thus the simulations-- Combined reform scenario, where the above three based on assumed future values of input variables--to modifications take effect simultaneously. start in January 2005. The simulation period is five years. Reduction of contribution rate scenario, where We distinguish the following scenarios: the contribution rate is reduced XX times (from Baseline scenario: This scenario assumes histori- the current YY to ZZ percent). The reduction of the cal values to prevail in the indefinite future. In par- contribution rate from 4 to 1 percent is assumed ticular, it assumes that the following key characteristics below. and parameters remain at their 2004 levels (average Economic shock scenario, where the inflow into 2004 values for flows and December 2004 values for unemployment and the receipt of UI benefits are stocks): tripled over a three-year period (from January 2005 till December 2007), returning to the original base- Level and duration of benefits, as well as eligibility line level thereafter. conditions for their receipt Number, age, and gender structure of the benefit Simulation results recipients Below we present the evolution of the key UI sys- Distribution by the length of stay in unemployment tem variables for the 2004­2009 period as predicted of benefit recipients by our simulation model. The output variables are as Number and structure of contributors by age and follows: gender, as well as their wage structure Contribution rates. Contribution rate required (expressed as the per- centage of the current contribution rate): the rate Alternative scenarios: To examine changes that that would equate current revenues with current would occur if some of the key labor market character- expenditures. istics and UI parameters change, the model is simulated The index of benefits affordable: the level of benefits assuming alternative developments. These developments (expressed as the percentage of current expendi- reflect contemplated changes of the UI system as well as tures) that could be supported by current revenues changes in baseline circumstances, which allow evalu- (for example, the index of affordability of 1000 ating the long-term sustainability of the system. Note percent means that current revenues could finance that under the alternative scenarios, only one parameter a tenfold increase of expenditures). 4 Unemployment Insurance Simulation Model (UISIM) The index of UI fund reserves, showing the number Figure 2: Simulated financial performance of of months for which the current level of benefits the UI system for Country X could be paid by the reserve fund alone (calculated as the reserves divided by current monthly expen- Contributions required ditures). 60% Number of recipients of UI benefits. 50% Average level of UI benefits (in national currency units). 40% Average duration of receipt of UI benefits (in 30% months; calculated as the average duration of receipt 20% for those recipients whose benefits cease in a par- 10% ticular month). 0% Share of UI benefits in GDP. Year The simulation results summarizing the financial Bene ts a ordable 140% performance of the UI system are presented in Figure 120% 2, showing: the required contribution rate; benefits 100% affordable under the current contribution rate; and 80% the capacity of reserves to finance the current level of 60% expenditures. 40% The simulations show that the required contribu- 20% tion rate (that is, the one that equates current expendi- 0% tures under different scenarios with current revenues) is Year below 25 percent of its current rate for all single change scenarios, which increase the generosity and/or ease the Unemployment insurance fund reserves access to benefits, and that it increases to a maximum of 1600% 1400% 37 percent under the combined reform scenario. Under 1200% the shock scenario, the required contribution rate peaks 1000% at 41 percent of its current rate. Of course, a drastic, 800% fourfold reduction of the contribution rate under the 600% reduced contribution rate scenario brings the required 400% contribution rate much higher, to 82 percent--even so, 200% 0% such a reduction of the contribution rate still generates Year current surplus to the fund. A similar picture of com- Combined reform Base Increase of bene ts fortable excess revenues over expenditures emerges from Reduction of contributions Increase of duration by 100% Easing of eligibility the simulations of the benefits affordable under the cur- Shock rent contribution rate and from the simulations of the UI fund reserves. To better understand the evolution of expendi- tures, it is instructive to examine other simulation outcomes, in particular the number of recipients, the reform scenarios. Similarly, the average duration of level of benefits, the average duration of receipt of UI receipt of UI benefits increases under the increase of benefits, and benefit expenditures as a share of GDP duration and the combined reform scenarios (adjust- (Figure 3). ment paths are quite complex, reflecting changes of Except under the reduction of contributions sce- inflow into recipiency following the change), but not nario, the number of recipients increases under all other under other scenarios. Another look at the simulation scenarios. The average level of UI benefits changes only results is provided by the evolution of benefit expen- under the increase of benefit level and the combined ditures as a share of GDP. Under the baseline scenario, 5 Unemployment Insurance Simulation Model (UISIM) Figure 3: Select simulated outputs of the UI these expenditures remain constant at 0.05 percent of system for Country X GDP. Under alternative scenarios, this share increases above its value in the baseline scenario. Its highest value Number of recipients --0.14 percent of GDP--is reached under the combined 180 reform scenario 20 months after the start of the reform, 160 and stays at that level thereafter. 140 120 Logistical Information 100 housandsT 80 60 The model has been developed and is managed by 40 20 the Social Protection Unit of the Human Development 0 Network at the World Bank. Similar to the PROST Year model, attending UISIM training is required in order for the software to be released. The contact person is Average bene t and further information may be obtained from Milan 300 Vodopivec (HDNSP). 250 200 TRLM 150 100 50 0% Year Average duration of receipt 8 7 6 5 4 3 2 1 0 Year Share of bene ts in GDP 0.1% 0.1% 0.1% 0.1% 0.1% 0% 0% 0% Year Combined reform Base Increase of bene ts Reduction of contributions Increase of duration by 100% Easing of eligibility Shock 6 Unemployment Insurance Simulation Model (UISIM) Appendix insurance) system, such as National Employment Office and National Pension Fund. These agencies provide data about the participants in the UI system--the number, Description of Data Sources and structure, and wages of workers covered by (and/or Definition of Variables complying with) the system, as well as the number and structure of the recipients of unemployment benefits. A Sources of data National Employment Office also can provide informa- UISIM relies on two main types of data sources: tion about the parameters and rules of the UI system. labor market surveys and administrative records. Both sources typically exist in all countries. Labor market sur- Definitions of variables veys provide data about employment, unemployment, UISIM requires data of specific format as an input and wages. Employment data should preferably include to the model. The data refer to the territorial unit corre- both formal and informal sector workers, although only sponding to the unemployment insurance pooling level workers covered by the UI system are taken into account (usually a country, but also a city or municipality). The in the sections of the model pertaining to the UI system. reference age category below is 16­60 years; however, The other key data sources are records from the gov- this feature may be adapted in the simulation to reflect ernment agencies administrating the UI (or other social the respective UI system. Table 1: Definitions of variables used by UISIM model Variable name Definition Remark A. Labor Market Employment Number of employed persons, by gender and age (by yearly cohorts, 16­60) Both workers employed in the formal or informal sector should be included (estimates suffice). Employment ­ workers who are actual contributors to Number of actual contributors to UI fund, by gender and age (by yearly cohorts, Actual contributors are workers who are paying UI contribu- the UI fund 16­60) tions, and/or whose employers are paying contributions for them. If available, also obtain data for 2004. Unemployment Number of unemployed persons, by gender and age (by yearly cohorts, 16­60) Depending on the availability of data, registered unem- ployed could be used instead (persons registered as unem- ployed with Employment Bureaus). Average duration of unemployment spells of unem- Average duration of completed unemployment spells of unemployed, in Based on completed durations. ployed months by gender and age (by yearly cohorts, 16­60) Work experience Average duration of working experience (cumulative period a person was This variable may not be needed, depending on the eligibil- employed at a given age) of unemployment benefit recipients at the time ity conditions for UI benefits. they register with employment bureaus, in years by gender and age (by yearly cohorts, 16­60) B. Unemployment Insurance Average wage Average monthly gross wage of employed ­ actual contributors to UI fund, in local currency, by gender and age (by yearly cohorts, 16-60) Minimum wage Minimum wage in local currency Expenditures of UI fund for payment of UI benefits Expenditures of UI fund, total and breakdown by categories (payment of UI benefits, other expenditures) Revenues of UI fund from payment of UI contributions Total revenues of UI fund obtained from the payment of UI contributions Unemployment insurance recipients Number of unemployment benefit recipients by gender and age (by yearly cohorts, 16­60) (continued on next page) 7 Unemployment Insurance Simulation Model (UISIM) Table 1: Definitions of variables used by UISIM model (continued) Variable name Definition Remark Average duration of receipt of unemployment benefits Average duration of unemployment spells of unemployment benefit recipients, Based on completed durations (tip: calculate the average by gender and age (by yearly cohorts, 16-60) duration of unemployment spells of unemployment benefit recipients who exited unemployment in a certain period, preferably a year to avoid cyclical effects). Other income of UI fund Other revenues of UI fund (estimate of yearly other revenues) Main sources may be interest revenues and government subsidies. C. Other Variables Inflation rate Inflation rate (yearly level) Nominal interest rate Nominal annual interest rate ­ nominal annul rate of return on UI funds when invested in bonds or other financial instruments Gross Domestic Product Gross Domestic Product (yearly level) Population Number of persons living in the country (or unit of analysis), by age and gender (0-99) The findings, interpretations, and conclusions expressed herein are those of the author(s), and do not necessarily reflect the views of the International Bank for Reconstruction and Development / The World Bank and its affiliated organizations, or those of the Executive Directors of The World Bank or the governments they represent. 8