“W5... . IlllllllllllllllllllUllllIllllllfllllNlllllllHlllHlllllllW 3 1293 10377 9165 LIBRARY Michigan Suva my THREE ESSAYS ON THE EFFECT OF EXPERIENCE RATING IN UNEMPLOYMENT INSURANCE By Terrence Charles Halpin A DISSERTATION Submitted to Michigan State University in partial fulfillment of the requirements for the degree of DOCTOR OF PHILOSOPHY Department of Economics 1978 ABSTRACT THREE ESSAYS ON THE EFFECT OF EXPERIENCE RATING IN UNEMPLOYMENT INSURANCE by Terrence C. Halpin The 1932 Wisconsin Unemployment Compensation Act called for an experience rated unemployment insurance (UI) tax, one which ties an employer's tax rate to his past employment ex- perience. Since that original UI statute, experience rating has gained favor primarily on the basis of three arguments. First, by making unemployment costs a cost of production firms should be induced to stabilize their own employment fluctuations in an attempt to minimize their production costs. Second, experience rating should encourage firms to participate in the administration and development of the UI system. And last, experience rating should lead to an optimal resource allocation since it charges the firms which create unemployment with the cost of that unemployment. This study is in the form of three essays. Each one looks at some facet of each of the arguments to see if any of them are valid. The degree of experience rating, i.e., how close firms come to balancing tax payments with expected benefit payments to ex-employees, can differ markedly from state to state because of tax rate ceilings and floors. These differences in degree Terrence Charles Halpin allow a test of experience rating's influence on firm behavior. The parameters of the experience rated UI system, that is the tax rates and the taxable wage base, are used to test the effects of rating. If experience rating the UI tax induces firms to stabilize, then states with strong experience rating should exhibit smoother employment fluctuations than states with weak. The first essay tests this hypothesis by looking at the seasonality of employ- ment across states in three industries, SIC 161, highway and street construction except elevated highways; SIC 233, women's, misses', and Junior's outerwear; and SIC 2H3, millwork, plywood, veneer, and other prefabricated wood products. A measure of seasonality is created by estimating the spectrum of the time series of employment for each industry for each state. Monthly employment data fron11960-197u are used. A linear regression equation is estimated with the spectral measure of seasonality as the dependent variable and the parameters of experience rating as the regressors. The results show that firms do indeed respond to experience rating by stabilizing their seasonal, at least, employment fluctuations. The second essay uses employer challenges of decisions to award benefits to separated workers as a proxy for employer interest and participation in the UI system. Other factors which might influence litigiousness besides the strength of experience rating are included. Cross section data by state for the years 1969-197“ are used. The effect of experience rating is tested in a regression analysis framework on the Terrence Charles Halpin pooled time series and cross section data set. The results show that several factors influence a firms propensity to appeal benefit awards, and that the strength of experience rating is among them. The final essay shows that if some unemployment costs are not due to any individual firm, but are socialized unemployment costs, then experience rating may not lead to an optimal alloca- tion of unemployment costs, and hence may not lead to an optimal resource allocation. Given some logical assumptions, the analysis shows that a mixed tax plan may be the best way to levy the UI tax with a flat rate tax for the socialized costs and an experience rated tax for the costs due to individual firms. The conclusion drawn from the three studies is that ex- perience rating should be strengthened. As a result decreases in unemployment can be realized in two ways: by dampened seasonal fluctuations, and second, by allocating resources away from unstable employment industries into stable employment industries. Both goals can be reached without employers becoming overly strict in their challenges of benefit award decisions. ACKNOWLEDGEMENTS Although my name alone appears as the author of this piece, many others have helped directly and indirectly in its develop- ment and completion. Special thanks first go to my mentor, Dan Hamermesh, who has helped me so much in my slow but sure, I think, development as an economist. I thank Ken Boyer, Subbiah Kannappan, and Dan Saks for their comments and guidance in this project. The companionship of good friends make any undertaking easier, and no one could ever have better friends than Evelyn and Mark Fallek, Joe and Crystal Stone, Mark Machina, Patty Bachmeier, and Laura Littleford. My appreciation and sympathy go out to Sandi Neubauer who typed this thing three times now and never once fell to sleep at the typewriter. I thank Kathy Halpin, whose intelligence and common sense helped direct and redirect ny analysis many times. Finally, I wish to thank Kathy again and Sarah Halpin for their love and understanding and especially their uncanny ability to always make me laugh even when this dissertation seemed like it would never end. 11 Chapter I Chapter II Chapter III Chapter IV TABLE OF CONTENTS Experience Rating and the Unemployment Insurance System Mechanics of Experience Rating Extent of Experience Rating Merits of Experience Rating 1. Stabilization of Employment 2. Employer Interest and Participation 3. Allocation of Unemployment Costs Outline of Study Effect of Unemployment Insurance on Seasonal Fluctuations in Employment The Unemployment Tax as an Inducement .to Smooth Employment Fluctuations Data and Methodology Regression Results Potential Problems Conclusions The Effect of Experience Rating on Employer Appeals The Appeals Process Theory Data and Methodology Regression Results Conclusions Experience Rating and the Allocation of Unemployment Insurance Costs Experience Rating and Cost Allocation The Model Experience Rating, Uniform Taxes, and Social Unemployment Costs Combination Tax Schemes The Optimal Amount of Experience Rating Conclusions 111 17 18 22 28 40 AZ 77 78 82 88 97 100 103 Page Chapter V Conclusions 105 Review of Findings 106 Recommendations for Change 110 Appendix Spectral Measure of Seasonality 113 Bibliography 116 iv LIST OF TABLES Page Table l Unemployment Insurance Tax Provisions, 1975 5 Table 2 Effect of Experience Rating Indicators on Seasonality A 30 Table 3A Regression on Split Industry 161 Sample 35 Table 3B Regression on Split Industry 233 Sample 36 Table 3C Effect of Experience Rating Indicators on Seasonality Split Industry 243 Sample 37 Table A Chow Tests for Split Samples 38 Table 5 Average Total Benefit Payments, Selected States H8 Table 6 Lower Authority Appeals, Illinois vs. Rest of 0.3. 56 Table 7 Appeals Based on Not Able and Not Available, Illinois vs. Next Highest State 58 Table 8 Chow Tests 60 Table 9 Regression Results 65 Table 10 Total Effect of AL and PMAX 70 LIST OF FIGURES Page Figure 1 102 Figure 2 103 vi CHAPTER I Experience Rating and the Unemployment Insurance System The unemployment insurance system in the United States is an amalgam of a federal program and 51 separate, and often dis- parate, state programs. Unlike Europe, where even several cities had unemployment insurance (UI) plans as early as the 1890's, in the U.S. it was not until the Social Security Act of 1935 that UI was established nationwide.l Prior to 1935, however, six states had enacted UI laws of their own. The pioneer in the field was the Wisconsin Act, adopted in 1932, whose philo- sophy and mechanics influenced the federal statute. The Wisconsin provisions for temporary benefits to workers unemployed through no fault of their own were similar to the European predecessors, but the financing of benefits differed significantly. The European plans required contributions from workers, employers, and government. The Wisconsin plan called for employer payments alone, and each employer's contribution rate was related to the amount of benefits drawn by his ex-employees. This linking of the contribution rate to a firm's employment experience has come to be known as experience rating. lSee Nelson [16] for a detailed early history of unemployment insurance in the U.S. Although the Social Security Act of 1935 mandated a federal unemployment insurance system, it gave each state the option of setting up its own 01 program. In fact it encouraged the states to do so by allowing a portion of the Federal tax to be excused in lieu of payments into a state fund. Given the choice between collecting taxes itself or seeing them whisked away to Washington, not surprisingly every state and the District of Columbia began its own UI program by 1937. Besides motivating states to establish their own programs, the Act encouraged them to experience rate their unemployment insurance taxes. Experience rating makes the individual firm's tax rate dependent on its past employment experience. A firm which has experienced few layoffs in a qualifying period, and hence which has generated few benefit payments, is taxed at a lower rate than a firm which has generated many benefit payments. As of January, 1978, the federal UI tax is levied at a rate of 3.“% on the first $6,000 of each covered worker's wages. The tax is payable entirely by the employer. By paying taxes into its own state's UI fund, a firm can credit those payments to its federal tax, potentially reducing the federal tax to 0.7% of its taxable payroll. Furthermore if the state experience rates its UI tax, the firm can still claim the full federal tax offset even if it pays its state tax at a rate less than 2.7%. Thus, for example, a firm with few layoffs that is charged state taxes at a 0.1% rate, still receives credit for paying at a rate of 2.7%, and pays a total UI tax of only 0.8% (0.7% to the federal government and 0.1% to the state fund). As a result of this bonus offset provision all states have adopted experience rated tax schedules. The federal government, out of the 0.7% on taxable payrolls that it retains, completely reimburses the states for the administrative costs of their own employment service commissions, enforces a few federal standards, and provides advice to the states. Benefit levels, eligibility rules, benefit duration, and tax rates charged are all up to the individual states. Mechanics of Experience Rating The original unemployment insurance law created an in- dividual UI account for each firm in the state. Money flowed into the account from the firm alone, and the individual account was the only source of money for benefits to the firm's ex- employees. If the firm's account was depleted, no further bene— fits were paid to its laid off workers. To increase the insur- ance factor and to protect workers, all states eventually adopted some form of a pooled fund. Now tax rates are determined as if each firm had an individual account, but the taxes are paid into a general fund from which benefits are withdrawn. Several formulas are used by various states to experience rate the tax, the most popular (32 states) being the reserve ratio formula.2 Each employer has an account to which his tax payments are credited and from which benefits drawn by his ex- employees are debited. The reserve ratio is the ratio of the 28cc Haber and Murray [9] for a discussion of the reserve ratio and other experience rating formulas. firm's balance (taxes minus benefits) to its annual payroll. The annual payroll is usually the average payroll for the prior three years. The firm's tax rate is adjusted to make the re- serve ratio equal to some given number. Actually the adjustment is such that the reserve ratio is within some range. When the ratio reaches the maximum, a reduced tax rate is assigned; if the ratio reaches the minimum, the tax rate is increased. Ad- Justments are usually made on an annual basis. Although the intent of experience rating is to balance tax payments with expected benefit withdrawals, some states have instituted minimum and maximum tax rates which make balanc- ing impossible. Thirty-eight states have nonezero minimum tax rates (see Table 1). Firms with no layoffs in these states cannot possibly balance their accounts since their tax payments necessarily outweigh their zero withdrawals. Most states also use multiple tax schedules, a lower schedule in effect when the state fund is of some given size, a higher one when the fund is depleted to some specified level. Thus, even though a state can possibly award a zero tax rate, it may not. Table 1 shows that in 1975 only 6 states awarded zero tax rates, and the minimum rate was as high as 3.9% (in Massachusetts). Tax rate ceilings may make it impossible for some firms to pay as much into the UI fund as its ex-employees withdraw. Table 1 shows that maximum rates charged in 1975 ranged from 2.7% in six states to 6.6% in Michigan. Benefits charged to a firm but not covered by that firm's tax payments are referred to as ineffectively charged benefits. In 1967 18% of the covered H.m H.H H.m m. mcmucoz w.m o H.: o szommHz >.m :. ~.m o HmmHmmHmmHz oom.z o.m m. o.m H muommccHz o.w A. m.m o escapee: H.m m.m H.m m. muummszommmmz w.m p. w.m H. ccszpmz o.m :.m o.m m. mchz o.m o.H o.m H. wcmesoq m.m H. m.: H. axosucmx w.m o w.m o mmmcmx 0.: o o.: o mon H.m mo. H.m mo. acmHUCH 0.: H. o.: H. mHOCHHHH 5.: H.H H.m m. onmpH com.» o.m o.m o.m m. HHmzm: wm.m mo. m.= :mo. «Hmpomu m.: H. m.e o eoficoam N..N m.N New H. .O.Q m.= m.H m.= H. mamzmHmQ o.m m.H m.m H. usoHpooccoo w.m o m.m o oomLoHoo H.: A.H H.: H. cachoeefico m.: m. N.: o mmmcmxp< m.m H. m.m H. mcoan< ooo.oa H.= m.H m.m m. memefle ~.m m. m.m m. msman¢ mama mama oaseccom oHceocom chosm mmchs .cowpmno .oompmno oHnmgo>mm mHnmpo>mm oomzav ommm xme xme ammoq xwa pmoz xmB owmz mHnmxmB Esemez EBEHCH: EsEmez EschH: mme .mconH>opm xme mocmhsmcH acoEHOHQEocs H mam<9 .oamH .m szh =.m3mq mocmpsmCH acoEmoHQEoca oumum no mconH>opm pcmoHuchHm= anmH .mpmscmh .mzmq mocmpsmcH psoENonEoca madam mm :omemmaoo "moonsom 0:.m ma. ~.N 0 wcHEozz 5.: 0 :.: 0 :Hm:OomH3 m.m 0 m.m 0 chprH> who: 000.0 0.m 0.m 0.m IIII coumcHnmmz h.m m0. ~.N m0. chHwLH> 0.m 0.H 0.m H. pCOEpm> a.m =.H a.m a. he»: 0.: H. III H. mmxme 0.: :. 0.: m. ommmoccoe ~.m 0 H.: 0 wuoxmm 30300 H.: mm. H.: mw. mcHHoLmo gusom 0.m N.m 0.: 0.H UCMHmH mnocm 0.: 0.H 0.: m. chm>H>mccom ooo.m mmm.m =NN.H m.m m. cowcho a.m e. a.m m. eeocmflxo m.m H. m.: 0 case m.= m. «.3 m. euoxeo cahoz >.: m. a.: H. mcHHOAmo cupoz 0.m m.H m.m m. xhow 3oz 0.m 0. 0.m H. oOmez 3oz 000.: m.0 N.H m.0 :. hmmpmh 3oz o.: a.m m.= mac. openness: :cz 0.m m. w.m m. acm>mz a.m H. a.m IIII mxmmpnmz mbmH mme mHzcmcom mHsconom Aczonm mmoHcs .0mwpm20 .mowpmno mHQMpo>mm oHnmpo>wm 00m:av mmmm . awe xme pmmmq was who: use mwma mHnwxM% Esstmz Esechzt Esstmz EseHcHz AcoscHucoov H mqm<5 firms in California had negative balances for the year amount- ing to $92 million. Accumulated negative balances for the state fund were $56k million. Most states carry forward negative balances, although some intermittently excuse them and allow the firms to start with a clean slate. States have instituted ceiling and floor tax rates for political and economic reasons. Ceiling rates "protect cyclically sensitive industries from bearing the brunt of a general busi- ness slump."3 They lessen the fear that unlimited liability for unemployment costs might deter hiring or might drive firms out of business. They enhance the insurance aSpect of U1 since a firm's past experience may not be an accurate indicator of its future layoffs. Non-zero minimum rates have been adopted under the assumption that UI is a social insurance program and there- fore sharing of some of the costs by all firms is appropriate.u Besides ineffectively charging some benefits, some states also noncharge benefits. Noncharging means that benefits are paid to an unemployed worker, but they are not charged against any firm's account. The costs of these noncharged benefits are somehow distributed among all the firms. Benefits are non- charged for a variety of reasons, but most noncharges are con- nected with disqualifications. A claimant found ineligible for benefits can in many states collect benefits after waiting some number of weeks. Also some states automatically noncharge 3Warden [21], p. 89. “Becker [A], p. 20. benefits if the worker was separated under potentially dis- qualifying circumstances.5 0f fifteen states surveyed by Becker6 in 1966, noncharged benefits accounted for over 12% of benefits in nine of the states. South Carolina in 1967 noncharged "7% of all benefits paid in the state. Extent of Experience Rating Because tax rate ceilings and floors differ markedly across states and because of noncharged benefits, the degree of experience rating, i.e., how close firms come to balancing their tax payments with expected withdrawals, can differ mark- edly. Table 1 shows the minimum and maximum rates possible in each state in 1975, and the minimum and maximum actually charged. Some states, like Michigan, Alaska, New Jersey, and Florida have wide ranges in possible tax rates. Others like Virginia, Utah, and Arizona have relatively narrow ranges. The differ- ences among states are more pronounced in actual rates charged in 1975. Thus Massachusetts, despite a potential range of 0.5 to 5.1 percent, charged at rates between 3.9 and 5.1 percent. Michigan, on the other hand charged between 0.7 and 6.6 percent, out of a potential 0 to 6.6. The strength of experience rating in a state's UI tax can be indicated by the tax rates it charges and the tax base it uses. The higher the minimum tax rate, the harder it is for stable employment firms to balance their 5Becker [A], p. 36. 6Becker [A], p. 105. accounts. The lower the maximum tax rate, the harder it is for unstable employment firms to pay enough taxes to equal their benefit charges. Each state is also free to set its own taxable wage base (at least $6,000, as of January, 1978) and some, like Alaska at $10,000, have chosen bases higher than the minimum. Once the taxable wage base has been reached, further wages paid to the worker are not subject to experience rating at all. Therefore, the higher the wage base, the stronger the state's experience rating. Table 1 indicates which states in 1975 had taxable wage bases greater than $6000. Merits of Experience Rating The virtues and drawbacks of experience rating have been debated in the U.S. almost as long as the very idea of unem- ployment insurance.7 Three arguments are usually cited in favor of an experience rated unemployment insurance tax. Differentiating tax rates according to employment experience encourages individual firms to stabilize their own employment. Second, an experience rated tax promotes employer interest and participation in the administration and development of its own state's unemployment insurance system. Third, an exper- ience rated tax allocates the cost of unemployment benefits better than a flat rate tax because it relates unemployment costs to the product that generated the costs.8 7Becker [4], p. 5-16 outlines the history and early develop- ment of experience rating. 83cc Haber and Murray [9], pp. 337-3u6. 10 1. Stabilization of Employment The pioneer Wisconsin unemployment insurance plan gained acceptance largely because of the claim that the program would encourage firms to smooth irregular employment patterns. An ex- perience rated unemployment insurance tax charges each firm with the cost of unemployment benefits drawn by its ex-employees. Thus the firm's unstable employment becomes a cost of production, to be minimized along with all its other costs. For the profit maximizing firm, minimizing its unemployment insurance tax bur- den is tantamount to reducing its employment fluctuations. The firm will stabilize employment as long as the marginal cost of a layoff (in terms of expected UI tax charges) outweighs the margi- nal wage bill reduction. Paul Raushenbush, who helped draft the Wisconsin Act noted: Just as employers are now required by law to pay workmen's compensation for accidents, so they should in future pay limited unemployment compensation to laid off workers. Both accidents and unemployment are industrial hazards and genuine production costs to be prevented where possible, but compensated for where unavoidable.9 Opponents of experience rating point out that the firm's opportunities for stabilizing its employment are for the most part limited to smoothing seasonal peaks and valleys. Frictional, cyclical, and technological unemployment are almost completely beyond the firm's control. And even seasonal employment stabili- zation is limited by the weather and festival days, factors which the employer cannot alter.10 P. L. Rainwater, former administrator of the Mississippi UI system, claims: 9A3 quoted in Becker [A], p. 52. lOSee Haber and Murray [9]. pp. 339-3u0; Warden [21], pp. 81-83. ll ’ghe most probable result of merit rating in operation will be a rate structure which will impose low rates on all employers who participate in the production of goods for a relatively stable market, high rates on all employers who take part in the production of goods for a market subject to severe fluctuations...The re- ward of rate reduction will go to the employers who have done nothing to earn it. The penalty of high rates will be imposed upon employers who are not so much inefficient as unlucky. Thus merit rating will bear little or no relation to merit. Their determination will depe d less upon good management than upon good fortune./F ' Hence the firm's tax rate is primarily determined not by its stabilizing behavior, but by the inherent stability or in- stability of its industry, 2. Employer Interest and Participation Proponents of experience rating claim that rating increases employer interest and participation in the UI system. Specifi- cally the firm should be induced to police the system, scrutiniz- ing the claims filed by its ex-employees to make sure that they truly qualify for benefits. Each state has established an appeals process so that employers can challenge benefit awards on various grounds, such as employment being terminated because of misconduct, the unemployed worker not actively seeking new work, and other grounds. Experience rating encourages firms to use the appeals process since, if the appeals board rules in the firm's favor, any benefits paid for that claim are not charged against the firm, and hence do not adversely affect the firm's tax rate. Without an experience rated tax, the potential gain 11 See Rainwater [18], p. 760. 12 from a challenged claim is split among all firms, thus greatly diminishing the individual firm's motivation to file an appeal. Critics charge that the increase in employer participation as a result of experience rating is insignificant,12 and that rating has caused the primary function of 01, which is providing benefits to the unemployed, to be subordinated to experience rating itself. Considering the objectives of unemployment compensa- tion it would seem those administering the program would be cheifly concerned with the payment of benefits. But benefits are not the major issue. Today most unem- ployment compensation problems at both the legislative and administrative levels are chiefly involved with, if not subordinated to, experience rating. Proposals to extend coverage, to liberalize benefits, to make determinations of seasonal industries--these and similar measures are tested for their effect upon a possible reduction in the employer's tax rate through experience rating. In short, employers and many administrators in their preoccupation with experience rating are con- verting the present system into a tax program rather than one designed to pay benefits to unemployed workers.13 3. Allocation of Unemployment Costs Varying tax rates according to employment experience, forces the firm to consider UI benefit costs as a cost of pro- duction. Those benefit costs which cannot be eliminated by employment stabilization are incorporated into product price by the competitive firm. Thus consumers of goods whose produc- tion generates unemployment pay the price of that unemployment. For example, consumers of highly fashionable women's shoes, 12Becker [A], p. 127. 13Clinton Spivey, as quoted in Haber and Murray [9], p. 3A6. 13 having seasonal production schedules and high unemployment, pay for the luxury of the seasonality if the tax is experience rated. A flat rate tax requires firms with below average unemploy- ment partially to pay benefit costs generated by firms with above average unemployment. As a result, products from firms with stable employment are priced too high relative to products from firms with unstable employment. Consumers buy more than the socially optimal amount of the unstable employment goods because of their artificially low price, and resources are mis- allocated away from stable industries and into unstable in- dustries. The allocative argument in favor of experience rating is independent of any employment stabilization gains that may be realized as a result of rating. Edwin Witte, one of the authors of the Social Security Act said: There are good theoretical arguments on both sides of this question (whether to experience rate the tax). To me the weight of the theoretical arguments seems to be in favor of experience rating. I hold his view not so much because I believe the experience rating will op- erate to reduce unemployment, but because I believe that the adjustment of contribution rates to the varying risks of unemployment of different industries and establishments is in accord with prevailing American concepts. In all other "insurances" in this country, rates are adjusted to the risk. Our private economy is grounded upon the concept that each industry should stand on its own feet. Honest cost accounting requires that all costs be ascertained and properly allocated to the commodities produced or services rendered. An industry which Operates intermittently occasions great costs to its employees and to society through its methods of operation. Whether it can or cannot operate more regularly, the unemployment which arises by reason of its intermittent or irregular operation is a cost which should be charged to the establishment producing the goods or services and which gets the profits of the enterprise. Every reason that can be advanced for con- tributions from employers only--and in all but six 1“ states all contributions come from the employers-~logically leads to variable contribution rates, rates adjusted to risk and costs. In a socialistic economy it might be proper to have all industry collectively bear the costs of unemployment; in a private economy, where the pro- fits go to particular entrepreneurs, all costs of pro- duction should be borne by the particular establishments, and these should include the unemplgyment compensation costs, as well as all other costs.1 (/ Opponents of experience rating maintain that not all un- employment is caused by an individual firm or by the production of a particular product (like fashionable women's shoes). For example, cyclical unemployment is associated with a general economic decline where cutbacks in one industry are often the result of cutbacks in another. It is rarely possible to lay blame for unemployment on any one industry, let alone on any one firm.15 Yet experience rating regards the firm which actually lays off the worker as solely responsible for the subsequent benefits drawn. Haber and Murray claim that "there are practical limits to how high maximum tax rates can be pushed."16 The "socialized" costs arising from ineffectively charged benefits and other costs ’"that cannot or should not be charged to any one employer"17 should be shared by all employers. Experience rating shares socialized unemployment costs only if each firm's share is com- pletely reflected in its layoffs.) l“Lampman [153. pp. 27A-275. 15warden [21], pp. 81-82. l6Haber and Murray [9]. p. 353. l7Haber and Murray [9], p. 357. 15 Outline of Study Despite the long standing political debate, little has been done to analyze the impact of an experience rated UI tax on the firm in a microeconomic framework. The purpose of this study is not to evaluate the arguments for and against exper- ience rating, but to investigate at least some facet of each of the three arguments to see if any of them are valid. Does experience rating actually induce firms to stabilize their own employment? Does experience rating encourage firms to scrutinize benefit claims more diligently? Does experience rating better allocate the cost of unemployment insurance than a flat tax rate? Whether the costs of unemployment insurance taxes are signi- ficant enough for firms to alter their employment fluctuations is an empirical question. The first section of this study considers the seasonality of employment across states for three industries: SIC 161, highway and street construction except elevated highways; SIC 233, women's, misses', and juniors' outer- wear; and SIC 2H3, millwork, plywood, veneer, and other pre- fabricated wood products. A regression model is used to see if an industry's seasonal fluctuations in employment are in- fluenced by the strength of experience rating. The second section uses employer appeals of benefit awards to ex-employees as a proxy for employer interest and partici- pation in the UI system. Again, whether UI costs are significant enough to alter firm behavior is an empirical question. A 16 regression model tests whether the proportion of benefit awards appealed by employers in a state is affected by the strength of experience rating in that state. The third section deals with experience rating as an allo- cator of unemployment insurance costs. The portion of U1 costs which are "socialized" costs, and therefore, for efficiency, should not be charged against any individual firm, is an open question. As a result, a theoretical approach is taken in this section, investigating the effect of experience rating on re- source allocation in the presence of different types of "social- ized" unemployment costs. The effect on resource allocation of an experience rated tax or a flat rate tax is considered in a situation where blame for a part of unemployment cannot be observed. The effect of the two schemes, and a mixture of the two, on resource allocation in the presence of "socialized" unemployment as a cost of maintaining the labor market, is also considered. CHAPTER II The Effect of Unemployment Insurance on SeasonaI FIuctuatIEns in Employment The possible stabilizing effects of unemployment insurance have been of interest to policy makers ever since the pioneer unemployment compensation law was enacted in 1932. At the bidding of gubernatorial candidate Phil LaFollette, that act was drawn up "with maximum emphasis on steadier jobs."18 When President Franklin Roosevelt called for UI to be included as part of the Social Security Act he asked for a plan whose goal was "to'afford every practicable aid and incentive toward the larger purpose of employment stabilization."19 Despite this longstanding interest and the arguments about the existence of stabilizing effects, little empirical work has been done on the subject. That which has been done deals ex- clusively with the macro stabilizing effects,20 ignoring the possible micro employment effects which had much to do with making UI acceptable to representatives some #5 years ago. 18 19 Raushenbush [19], p. 18. Nelson [16], p. 212. 208cc von Furstenberg [20] and Hamermesh [10] for recent macro stabilization studies. 17 18 The bulk of micro empirical work concerning UI has involved the effect of benefits on supply side behavior in the labor market.21 This study investigates differences in demand side behavior caused by the manner in which the UI system is financed, through an experience rated tax on employers. Employment stab- ilization is gained at the firm level by experience rating the UI tax, that is, tying the individual firm's tax rate to its own past employment experience. By making the tax payable by the employer, it becomes a cost of production. The profit maximizing firm, in seeking to minimize its tax burden along with its other costs, stabilizes its own labor requirements.22 The Unemployment Tax as an Inducement 22 Smooth EmpIOyment FIEctuations Unemployment insurance programs in the U.S. are a pre- dominantly state run endeavor. The federal government sets a few standards, provides advice, and reimburses the states for the administrative costs of their systems. Regular bene- fits are financed by state UI taxes. Only three states (Alabama, Alaska, and New Jersey) call for employee contribu- tions; all others tax the employer alone. 21For a summary of empirical research see Welch [22]. 22Brechling [5] develops a more rigorous theoretical model of how experience rating can affect firm behavior. 19 A state can differentiate tax rates among firms by exper- ience rating its tax. An experience rated UI tax is one in which the tax rate assigned to a firm is dependent on that firm's past employment experience. A firm which has generated a large number of layoffs in the qualifying period, and hence has generated a large amount of benefit claims, is charged at a higher tax rate than a firm with few benefits charged against its account. Thus each firm is in a manner paying for the unemployment that it creates. A perfectly experience rated tax charges each firm whatever tax necessary to balance its payments into the UI fund with the expected withdrawals from the fund by its ex-employees. A firm with no layoffs would be assigned a zero tax rate, a firm with many layoffs assigned a rate high enough to ensure balanc- ing. In fact, however, states have statutory minimum rate floors, maximum rate ceilings, and taxable wage bases less than total covered wages, so that balancing of accounts does not always occur. The degree of experience rating, i.e., how close firms come to balancing their accounts, can vary markedly across states. These differences in degree allow a test of experience rating's ability to smooth employment. The stronger the experience rating in a state, the greater should be the motivation to smooth employment since the stronger the rating the more are unemployment costs coming home to roost. Thus states with strong experience rating should exhibit smoother employment 2O patterns for a given industry than states with weak experience rating, all other things being equal.23 The contention that experience rating can induce firms to smooth their employment fluctuations assumes that the firms can alter those fluctuations in some way. A firm might stabilize its employment by reworking production schedules or by taking on complementary lines to reduce the seasonality of its employ- ment. Clearly part of the firm's employment variation is beyond its control. Things like seasonality of demand for the firm's product, weather considerations, and seasonality in raw materials supplied to the firm greatly limit the amount of stabilization actually possible. Whatever employment variability that the firm can affect, however, should be dampened by exper- ience rating the unemployment insurance tax. Experience rating's ability to induce firms to smooth employment is also limited by the amount of the tax that the firm can shift either backward onto labor or forward onto con- sumers. If the firm draws its workers from a wage inelastic labor supply, it ought to be able to shift all of the tax costs 23See Becker [A] for a more detailed account of the link between experience rating and employment stabilization. He reviews three prior studies, each of which deal with firm re- action to experience rating within one state. The studies all employ interviews or questionnaires and find that, in general, about one fourth of the firms sampled (having one third the covered workers) are appreciably affected by experience rating incentives. Feldstein E7] develops a model showing that unemployment insurance increases temporary layoff unemployment. Imperfect experience rating is partially responsible for the increase in that it subsidizes firms with poor employment records such that they do not pay the full cost of benefits drawn by their laid off workers. 21 back onto its labor force. Similarly if demand for the firm's output is price inelastic it should be able to shift forward the entire tax. This shifting problem is ignored here; i.e., it is assumed that the firm bears the full burden of the unem- ployment insurance tax. If shifting does occur it should weaken the incentive to smooth employment since it implies that the firm is escaping part of the tax. If experience rating induces firms to smooth their employ- ment swings, then a given industry should exhibit smoother swings in a state with strong experience rating than in a state with weak rating. Specifically this paper looks at seasonal swings in employment. The reasons for this are twofold. First, theoretically, if this smoothing behavior is going to show up anywhere, it ought to show up in seasonal variations since those' rather than cyclical or trend variations are more under the control of the firm. (That is, control in the sense that they are more predictable and hence more easily adapted to by the firm.) Secondly, for practical considerations, looking at seasonal variations allows one observed cycle for each year's worth of data, rather than, for example, one observed cycle for three years data if estimating a three-year business cycle.”4 Given some measure of seasonality for an industry for each state and given some parameters of experience rating strength for each state, then one can test to see if the parameters of 21‘See Hamermesh [11] for a brief discussion of the importance of a long time series. 22 strength of experience rating influence the seasonality of that industry in some characteristic manner. Data and Methodology Three industries were chosen for investigation. They were SIC 161, highway and street construction, except elevated highways; SIC 233, women's, misses', and juniors' outerwear; and SIC 2A3, millwork, plywood, veneer, and other prefabricated wood products. These were picked since they should exhibit seasonality, and they are also well represented through the fifty states, ensuring a good sample size. To get a measure of seasonality for each industry for each state the spectrum of a time series of employment for the industry was estimated. The spectrum shows how the variance of the time series is dis- tributed across frequencies. By totaling up the portion of variance accounted for by the seasonal frequency and its har- monics, a measure of seasonality for each industry for each state is obtained. Monthly employment observations by state25 for the years 1960 through 197“ were used to create the time series for each industry. Because of missing data or an industry not being represented in each state, the number of spectral measures of seasonality differed from industry to industry. Fifty spectra were estimated for industry 161, 36 for industry 233, and 39 for industry 2N3. Each of the time series was greatly dominated 25Data source: Employment and Wages [28]. 23 by low frequencies, making prewhitening of the series a must before estimating the spectra.26 The procedure used was to estimate a first-order autoregressive parameter for each series and to use that parameter to prewhiten the detrended series. Thus the actual spectra estimated for an industry were for the series zti - xt1 - axt_l, i’ t = 1,179; 2180, 1 = a(x179 - x0); where x is monthly employment, 1 indexes the states. The spectra were estimated using a Parzen window and 36 lags,27 so that the Spectrum for each industry for each state was M s(wJ) = l [10 C0 + 2 E Ak Ck COS W k] z; k-l 5 where M is the maximal lag (here M = 36) and W3 3 l%) J a O: M and where the estimated covariances Ck are given by [T'k < ‘ 1 C 8 1 X z - E) (z - z) k ‘T t=l t t+k ‘_ T where z I %’ X z and T is the length of the time series (here t=1 T a 180). The Parzen window is the weighting scheme - 6 k M 2 1k . l 660(1):: + (13,13)3 5' / “1-333 M/2oo Hmpou op memos oHnmxmu no OHuwm 909\x sconcoaoo mocoHaoaxm no vacuum .oocwpsmcH acoEHoHQEoCD 2H mcHuwm ooonemem .poxoom .z summon .mamH .coHumppchH804 wchHmeE1ccw ucoeonQEm .pommulmo pcosupwaoa .m.0 my .mamH .m mumscmh .mzma monogamCH unoanHQEocs opmum mo aconH>oem ucmonHcmHm .coHumauchH50< Lozoacws .aonmq mo wcosupmaoa . .0 .ame .momH .mmmH memo» pom .Npumsch no moumm xme oocmasncH psoEHOHmEocD .HF0H .conmnpchH50< nozomcwz .eonmq no ucoEupmmoa .m.0 .0amHI0mmH mama HaHocmch mocwnsmcH ucosmoHQEoCD "hoopsom .mewoz o>onw po>o commune mama EsstwE owwpo>< m< .mumoz o>onw po>o commune ouch EschHE owmuo>< A< .memoa o>oom Eoem commune when EsstmE unoan: x was» no ous> :choE on» .Ham. a Bos\xhomno mm mm: MHQEansm scam ammm. sm:m. 5 3 MH:m. amm:. Aem.H0 m:Ho. Adm.H0 same. m< mH.ou Ama.o0 Aoa.mu0 Adm.m0 amoo.u 0 mooo. m::.ma 00mm. Amm.H0 AeH.o:0 Azo.o:0 Amm.au0 Ama.m0 omao. Hmoo.u Hmoo.n same: mmam. eoexxee so: om.H Amm.ov Ham.~uv Amm.ev mmo. 0 eooo. Homm.u memm. Ama.H0 Aoo.m0 Aaa.ov “mo.mn0 Amm.e0 mmflo. mmmo. mooo. =~mm.: «mam. soE\x zuHHmcomwmm no chance: Hweuoomm ”oHanem> ucmpcoaoo oHQEmm HwH zhumsccH uHHmm co COHmmoLwom mm mqm<fi on» .mcoHum>eomno pH was azopw ann 028 .Ham. n eoexxee no pHHom .mH asopw 30H no: oHdEmm HmcHano one . AeH.m0 Amm.ov ommm mode. maoo. amam. ,o gem.mv :3 Hmmo. Amm.:0 Heoa. omeo. mmaa. mm me he H:0.Hv moHo. Am=.H0 mmoo. X huHHmcommom mo cashew: Happoomm mHQEmm mmm >Lum:ch uHHmm co mconmopwom am mqm ucoocodoa Amo.o0 Aam.ouv Aao.onv AHo.Hu0 Aaa.~0 0000. 0000. mm00.l H000.I 0mmH.| Hm0m. Amm.o-0 Amm.anv Amo.ov Amm.au0 Amo.m0 . 0000. :0mm.l 050m. 908\x<9 sod H0:H 7 3 Amm.H0 Hom.o-0 Amm.o0 Aam.o0 Ama.o0 00mm. aaHo. 0000.: 0000. 0m00. 00a0. Aom.on0 Ame.ov Am:.o0 A00.ov . moam. mooo. Hmao. memo. 909\x szH020nmom mo cashew: Hmpuooam ”oHanpm> unoccoaoo oHQEmm m:m heumsocH pHHQm apHHmcommom co mpouMochH mcHumm oocoHeoaxm mo pommmm om mamas ................. TABLE A Chow Tests for Split Samples S = H (C,TAX/TOT, PMAX, CAMIN, CAMAX) Industry 161 F a 1.65 (F a 2.00 at 10% level) Industry 233 F - 2.51 (F = 2.08 at 10% level) 3 = H (c, TAX/TOT, PMAX, AL, AH) Industry 161 F a 1.43 Industry 233 F = 2.69 industry faces its own limitations on its ability to stabilize, then it should react differently depending on its own powers of adaption, i.e., its own control over altering its labor require- ments. The construction industry SIC 161 is a highly seasonal one; in fact SIC 16 is the most seasonal industry in the two digit classification. These construction firms are almost al- ways in the maximum tax rate group, such that ranges of tax rates, minimum rates, and the like really should not influence them much since they are out of the realm of possibility. What can be of importance to these firms, and what the estimation procedure shows they react to, is the amount of wages on which they must pay the UI tax. The tax increases the cost of labor. So, for example, it is less likely that a worker expected to be needed for only a short period of time will be hired. Therefore, if construction firms attempt to queue their contracts rather than make use of a surplus labor supply, they may be doing so in re- sponse to the experience rated UI tax. 39 The outerwear industry SIC 233, on the other hand, reacts more to the tax-rate aspect of experience rating strength than to the base.. But this is a low wage industry such that more of the wages are within the wage base limit no matter how high or low the taxable wage base. Also this is an industry with enough control over its labor requirements that it can adjust more easily, so that tax rates other than the maximum are within its reaching distance. That this industry has more options for adapt- ing is supported by the fact that historically the garment in- dustry is one which has attempted to stretch its output over the year. Firms in the industry have wider options for adapting, eSpecially in that they exercise more control (predictability) over seasonal fluctuations in demand than a construction in- dustry. Given the greater stability in the seasonal fluctuations,' both in period and amplitude, this industry is in more of a long run situation than the construction industry, in the sense that its seasonality is more predictable, and hence one would expect the diminished effect of the taxable base variable. That firms in the same industry react differently depend- ing on the strength of the rating is further evidence that ex- perience rating is actually motivating firms to smooth their em- ployment. The outerwear industry shows significantly different firm behavior when split at the median taxable base level. Given the same type firm in two separate states, one with strong rating, the other with weak rating, the smoothing claim would say that the firm in the strong rating state should be more sensitive, and that contention is borne out. The construction A0 industry does not show different behavior that is statistically significant. But again it is one with little power to change its labor requirements, so that one would not definitely ex- pect different behavior within the industry. Also the choice of splitting the industry sample at the median TAX/TOT value was rather arbitrary. Given that SIC 161 comes close to showing a significant difference in behavior perhaps there might be some higher split point, i.e., some sufficiently strong degree of rating, which would show significantly different behavior. It should be noted that the assumption that firms cannot shift the UI tax is the worst case choice for finding experience rating to be signigicant. Any shifting of the tax burden diminishes the incentive to smooth employment, since then that portion of the cost of U1 is no longer borne by the firm. Hence for a given degree of experience rating strength, if shifting can occur, the observed smoothing behavior is at least as great as it would be without any shifting. The no shifting assumption was made simply because the available data are not detailed enough to make any definite judgement. Potential Problems There is a potential problem in drawing conclusions from the above results, namely one of pinning down causality. The estimated equations show correlation between the strength of experience rating and dampened employment swings, but does that imply that one is causing the other? Might it not be that seasonal industries locate in states with weak rating? What we are 41 investigating here, though, is firm behavior within an industry across states. 0f the three industries analyzed here only the outerwear industry SIC 233 has much opportunity for movement. The wood products industry cannot easily move away from its source of raw materials and the construction industry is closely linked to local markets. A three digit industry classification is a rather fine category breakdown, such that firms in this class ought to be fairly homogeneous. If the firms in this seasonal outerwear industry can simply pick up their sewing machines and move to states with weak experience rating, it seems unlikely that they should be represented in forty-six states in 1974, the last year in the sample, (especially given that those states exhibit) wide differences in experience rating strength). That the same type seasonal outerwear firm is able to sur- vive in a state with strong rating, ostensibly by smoothing its employment, when a weak rating locale is available, seems evidence for causality. That the same type firm acts differently across states in adjusting its employment swings to the strength of rating in the state seems strong evidence for causality. A simultaneity problem exists if the seasonality of em- ployment influences the strength of experience rating. It is politically easy for states with a high concentration of stable employment industries to have strong experience rating since no firms are heavily taxed in any case. The analysis here, again, deals with industries at the three digit level, so that the firms should be quite homogeneous. That the seasonality 42 of employment for the state can influence experience rating strength does not say that these individual firms considered can influence it. All we can say definitely now is that exper- ience rating does influence firm's stabilizing behavior. Conclusions What has been gathered here is, I think, some good cir- cumstantial evidence that experience rating does motivate employers to smooth their employment swings. Only three in- dustries were studied and only two showed any reaction at all. But the reaction was logical and statistically significant. This is not to say that this smoothing behavior justifies ex- perience rating. There are other experience rating effects which must be considered, and even within this smoothing aspect no question was raised as to the benefits gained from smoothing relative to the costs incurred by administering an experience rated system. The purpose of the study was to test, for the first time, whether rating strength at all influences employers to alter their employment patterns to smooth fluctuations. And the evidence seems to say that it does. CHAPTER III The Effect of Experience Rating on Employer Appeals Proponents of experience rating claim that an experience rated unemployment insurance tax should increase employer interest and participation in the UI system.31 In particular, firms should police the system by contesting benefit awards to ex-employees who do not truly qualify for benefits. Each state has an appeals process through which employers can con-‘ test, on several eligibility grounds, a decision to award benefits. If the appeals officer upholds a firm's challenge, any benefits paid are not charged against the firm's UI fund account. The propensity to appeal benefit awards depends on the expected loss from not appealing. In the case of an experience rated UI tax, that expected loss is an increased tax rate. Strong rating implies that the firm's tax rate will be adjusted to reflect the full cost of benefits paid. Weak rating implies that the firm may escape part or all of the benefit costs, depending on how sensitive the tax rate is to changes in the employer's UI account. 31Haber and Murray [9] summarize the objectives of experience rating, its historical development, and its characteristics. Becker [4] does likewise and includes several specific examples of the effect that experience rating has on the manner in which individual firms handle unemployment insurance claims. 43 44 Many other factors affect an employer's litigiousness, the cost of appealing in terms of time and money, the cost of moni- toring claims, the probability of success, and the expected benefit cost. Holding constant these other factors, the stronger the experience rating, the greater should be the firm's prof pensity to appeal, since strong rating increases the cost of a forgone appeal. Employer policing of the UI system can take forms other than monitoring benefit claims. For example, firms might pres- sure the state legislature to enact stricter eligibility standards, or call for stricter enforcement of present rules. This study limits itself to policing as reflected in firm monitoring of ex-employees' eligibility for benefits. The fraction of total claims which are contested by employers is a proxy for the diligence of this type of policing effort, and it is used to test whether experience rating has any effect in this area. The Appeals Process In 1974 18,865,179 persons filed unemployment insurance benefit claims. A case may be illustrative. Mr. A from Illinois, was laid off from his job as a press operator from June 26 until July 15 due to a plant shutdown. He intended to take a trip to Greece in the interim and requested extra time off. His employer refused the request and scheduled him for work on July 15. Mr. A took the trip anyway with the intention of 45 leaving Greece on the fifteenth. Unfortunately political unrest made his return impossible, and worse, Mr. A was inducted into the Greek army on July 21. Efforts to contact his employer were futile since all communi- cation out of Greece was out. When he finally returned to the U.S. Mr. A found he had been discharged as of July 19 for failure to report to work. He filed for UI benefits on August 5 to be effective September 22, and they were granted. His employer, however, challenged the award, claiming that Mr. A was discharged because of misconduct and hence was not entitled to benefits. The Illinois UI Board of Review agreed with the em- ployer, reasoning that Mr. A's plans to remain in Greece until the fifteenth showed a "disregard of his duties toward his employer and constituted misconduct."32 His application for benefits was refused. Thousands of cases like this involving misconduct, no doubt most of them more mundane, are heard every year in states across the U.S. They are heard not in courts, but as part of each state's unemployment insurance system. Each time a deci- sion is made whether to grant or not to grant unemployment insurance benefits, both the claimant and his former employer have recourse to dispute that determination. The process through which an application for unemployment benefits is accepted or rejected, and the provisions for 32See the Benefit Series Service Unemployment Insurance Report [27J, Report BIO-29,7March 1976. 46 subsequent appeals of that decision differ from state to state. Michigan's system is not atypical. Upon being separated from work an ex-employee can file for benefits at a local Michigan Employment Security Commission office. The office determines if he is covered by unemployment insurance, and if covered, whether he is eligible for benefits. In 1976 a Michigan worker must have been employed for 14 weeks by one or more qualifying employers and have received a minimum of $25.01 per week in wages. The reason for the worker's separation also affects his eligibility. The most common grounds for declaring a worker ineligible are separation for misconduct, voluntary separation without good cause, not able to and not available for work, and refusal of suitable work. Adverse eligibility rulings can cancel all or part of a claimant's benefits. Once the applicant's eligibility has been determined, his weekly benefit amount, and the potential duration of those benefits, is calculated. The MESC deputy's eligibility ruling can be challenged by both the ex-employee and by his former employer. Only the most recent employer can contest a decision to award benefits. The disputed claim case is heard before an MESC referee (some states use a panel consisting of several persons to hear challenges). The appeals referee serves as an impartial, quasi-judicial officer who conducts evidentiary hearings per- taining to the claimant's benefit rights and the liability of the employer. Each party testifies for the record. They may be represented by a lawyer or an authorized agent. The 47 referee's prime responsibility is to obtain all of the facts and to secure relevant documentary evidence in the form of exhibits.33 The referee's decision can be appealed to the three mem- ber Michigan Employment Security Appeal Board, and its decision can be appealed to the state courts. All but four states pro- vide for a higher level of appeals authority within the UI system. Most use either a three member panel as in Michigan, or the head of the state's employment security commission to hear appeals of lower authority decisions. Connecticut, Hawaii, Nebraska, and New Hampshire utilize only a one stage appeals process before sending contested claim cases to the state courts. Theory An employer's decision to contest an award of benefits to an ex-employee may be based on personal animosity or pecuniary gain. Since the UI tax is experience rated, it may be rational economic behavior for the firm to contest a benefit award. The greater the benefit drain on a firm's UI account, the greater the required tax payments, via a higher tax rate, on the part of the firm. Table 5 lists the average benefit payment for several selected states for the period 1969 through 1974. If the employer successfully contests a benefit award, then no benefits :re charged against his account, and his tax rate is 33See the Michigan Employment Security Commission Annual Report 1974 L23] for a more complete description of Michigan's UI system procedures. '148 .mcmu noHanpmum mocmszmcH psoENOHQEocD “season when :o.am0 mm.m00 :m.H:m 00.a00 ::.0:0 mH.0=m :HocoomHz mm.mmm mm.smm am.mom mm.mme om.mmm Ho.mom omcmmpms om.amm mm.=mm Ho.mmm mm.amm ma.mms em.mo: memos :a.:mHH m0.m0m m:.aHoH 00.0mm 0m.:ma mm.0m0 meow 3oz mm.mmm oo.omm oo.mmm mm.mmm am.sms mm.mme denotes: ms.mma mm.msa Hm.mma mm.s=m am.=ma ms.em= cheeses: mm.mmom mm.mmom ma.asm om.:mos om.mma ms.aom mopoosmoeoocz mm.msm Ho.amm mm.ssm mm.emm as.amm aa.eae oocposoo ma.mmm ms.mom Hm.ema a=.amm am.=ma ms.m=m eacsoamsco mm.mm= ms.mm= m=.mmm No.0mm mm.mm= ms.ame oeoocme mama mama mama Hams cams moms opeem m mmmms moumum pouomHom .mucoEmmm pHuocom Hmuoa mwwso>< 49 not adversely affected. In states with strong experience rating the representative firm pays a larger portion of the cost of its separated worker UI benefits than in states with weak rating. Hence, holding everything else constant, aggregate firm behavior in states with strong rating should exhibit more diligent scrutiny of benefit claims, than in states with weak exper- ience rating. As alluded to above, employer interest and participation in the UI system could extend to working with legislators and administrators to alter the system, rather than merely policing ex-employee benefit claims. What some consider undue emphasis on both aspects has led to stiff criticism of the experience rated tax scheme. Walter Reuther in testimony before Congress in 1965 said: We know the pitiful results which experience rates have produced in the form of vicious disqualifica- tion provisions, unreasonably high earnings require- ments, cancellation of benefit rights, and worst of all, the baseless unreasoning, devious contesting by employers of all claims which may adversely affect their tax rates. Every device imaginable is employed to beat down, wear out, and defeat the unemployed worker who has the temerity to file a claim for benefits. Trade associations encourage this process of contesting all claims. They point out that dollars are involved in a claim. And they say that no matter if the employer does know that the worker is entitled to benefits under the facts and the law, he should still contest the claim because that will delay things, cause the State agency to investigate and, maybe, cause the worker to give pp the whole thing as an impossible fight to win.3 3“See Becker [4], pp. 128-130 for Reuther's quote and a discussion of it and other similar complaints. 50 Andrew Biemiller of the AFL-CIO testified in 1970 concerning unemployment insurance: We are convinced experience rating has led to the development of most unfair and undesirable practices within the program. Harsh disqualification provi- sions in state laws, and unmerited employer challenges of legitimate claims in order to preserve favorable tax rates flow directly from resent experience rat- ing requirements in the law.3g Whether experience rating induces perverse behavior on the part of firms in the form of;challenging valid UI benefit claims is not the question in this paper. What is under in- vestigation is whether the effect of experience rating is significant enough to encourage firms to dispute proportionately more claim applications in states with strong rating than in states with weak rating. An employer's decision to appeal a benefit award, assum- ing that he is acting on economic reasons, turns on whether the expected gain from winning the appeal outweighs the cost of the appeal. As intimated above the stronger the state's experience rating, the greater is the expected gain from an appeal, holding everything else constant. If the experience rating were so weak that each firm pays the average tax rate, then the expected gain from the reversal of a benefit award is quite small to the appealing firm since the tax gain is diSpersed among all the firms in the state. If the state has perfect experience rating, such that the appealing firm is solely charged with the benefits paid, then the expected gain 35Quoted from Biemiller's U.S. Senate testimony [26]. 51 from a reversal accrues to the firm alone, and hence the re- versal has a much larger impact. Aside from the tax saving, the firm can gain from con- testing a claim through the deterrent effect that the challenge can have on future claims. A spurious claim contested today might preclude future spurious claims, thus saving the firm the cost of future appeals. Let G be the expected gain to be realized from contesting a claim. Then G = G(TS, D) where TS is the expected tax saving and D is the deterrent effect. TS is actually a function of the total expected bene- fit payment and the strength of experience rating, since rating determines how much of the benefit payment will be charged to the firm. More completely, then G = G(TS(EXBEN, R), D) L109 36:09 19,10, aEXBEN 3R am where EXBEN is the expected benefit payment and R is an index of experience rating strength. Contesting a claim is not a costless proposition for the firm. It must prepare its case and present it, each of which costs time and money. Because of the possibility of higher- 1evel appeals, the firm may be forced to argue the case several times. The size of the firm might affect the cost of disputing a benefit award. Firms with large work-forces might be able to 52 take advantage of economies of scale in monitoring benefit claims if a high volume permits the use of a person or divi- sion which specializes in applying the intricacies of the law to claims filed. Large firms may also be more concerned than small firms with the effect on future claims of allowing a current erroneous benefit award to go uncontested. On the other hand, there are reasons for small firms challenging more claims, so that the effect of firm size is ambiguous. The manager of a small firm is more likely to have first hand knowledge of the conditions under which an employee was released. Whether there are grounds to warrant a claim challenge can be easily determined. In the case of a large firm the claims monitor might be forced into the tedious process of finding out the name of the ex-employee's supervisor, why and under what conditions he was released, and perhaps personal information about the worker which may affect his claim status (medical considerations, for example). Also, for large firms to take advantage of the possible economies of scale in scrut- inizing claims they must first institute some sort of monitoring system. For a firm with a very stable employment record the expected gain to be realized from such a system may not be large enough to offset the cost of maintaining it. In states with weak experience rating the potential tax saving even for a firm with above average separations may not be great enough to offset the special monitoring costs, whereas in a state with strong rating the cost might be justified. Thus whether large firms take advantage of the possible economies of scale depends partly on the strength of experience rating. 53 Firms may be motivated to contest claims for other than economic reasons. When a worker is fired‘or quits a small firm because of a job-related dispute, that dispute is more likely to be with the same level of management that makes the decision to challenge a claim. As a result more challenges out of spite, out of a desire to get the last word through denying benefits, are likely in small firms. Also workers leaving small firms voluntarily might be perceived as disloyal if the small firm does not experience the same sort of turnover as a large one. The effect of firm size, then, is indeterminate. Let L be the expected cost of contesting a benefit award. Then L = L(SZ, HLA) 3L 3'0, ‘8L 1 0, asz > aHLA where 82 is the firm's size and HLA is the probability that a higher level appeal hearing will be required. The firm decides to contest a benefit award if G > L and is indifferent if G = L. Let P(A) be the probability that a given UI benefit award determination will be challenged. Then P(A) = F(G - L) where 3P(A) :0, aP(A) _>_o, 3P(A) _>_o, 3P(A) _<_ ——3EXBEN as ED 352 > Ta A 54 Data and Methodology The data used to test whether experience rating encourages firms to police the unemployment insurance system more dili- gently are aggregate state data for forty-nine states and the District of Columbia (see below why only 49 states were included) from the six-year time period 1969 through 1974. The estimat- ing equation is of the form P(A)it = F(Rlit, R21, EXBENit, szi, Kit) 1 = 1, 50; t = 1969, 1974, where P(A) is the number of lower authority employer appeals per 10,000 new spells of insured unemployment, R1 is a group of time-variant indicators of experience rating strength, R2 is a group of time-invariant indicators of experience rating strength, EXBEN is the expected total benefit payment. 82 is a group of indicators of firm size in the state, and X is a vector of control variables. Table 8 lists the particular forms of the estimating equation. The deterrent effect D is difficult to quantify and was dropped from the analysis. The deterrent effect itself should not differ greatly across states unless the labor force's psychology differs from state to state. The gain from the deterrence, however, does differ if UI costs differ across states. The effects of those cost differentials are controlled for in the regression model. Also the cost consideration of higher level appeals HLA was dropped, since data do not exist 55 for the fraction of employer challenges which are appealed to higher authorities. This effect should not be significant, since the major cost incurred is the initial preparation of the case, and since less than 17 percent of total lower author- ity appeals (of which employer appeals are only a small fraction) are appealed to the higher authority appeals board. Data from Illinois were dropped from the sample because of this state's extraordinary contested claims behavior. While the average fraction of contested claims for all UI systems other than Illinois was 3.90 per 10,000 claims over the time period, Illinois' rates ranged from 19.15 to 34.30 with an average of 27.01. The discrepancy can be attributed chiefly to the Illinois 01 administration's particularly strict inter- pretation of the "able and available for work" eligibility criterion. During the period of the study an Illinois employer needed only to assert that a separated worker was unable or unavailable for work. The burden of proof was on the ex-employee. As a result Illinois experienced an inordinately high percentage of contested claims. Table 6 outlines Illinois' lower authority appeals behavior versus the average for the rest of the na- tion. Table 7 shows the percentage of benefit cases contested on the not able and not available grounds versus the next highest state for the six years under consideration. Because of the particularly stringent eligibility requirement several large Illinois firms automatically contested any UI claim by a former employee. Because of these unusual circumstances (the Illinois eligibility requirements have since been changed) Illinois was dropped from the sample. 56 TABLE 6 Lower Authority Appeals, Illinois vs. Rest of U.S. Total Voluntary Misconduct Year Decigions # Quit ' # __z___ 1969 Illinois 16,629 2,646 15.9 1,572 9.5 U.S. 172,194 69,256 40.2 30,094 17.5 Illinois 20,809 3,570 17.2 2,439 11.7 1970 U.S. 214,821 89,461 41.6 41,972 19.5 Illinois 26,656 4,945 18.6 3,280 12.3 1971 U.S. 261,677 109,761 41.9 56,756 21.7 Illinois 25,669 5,232 20.4 3,151 12.3 1972 U.S. 283,837 119,511 42.1 58,121 20.5 Illinois 26,524 5,486 20.7 3.582 13.5 1973 U.S. 306,962 128,001 41.7 61,426 20.0 197” Illinois 32,950 5,970 18.1 4,350 13.2 U.S. 358,695 14,761 41.2 85,175 23.7 57 TABLE 6 (continued) ngggzbl: Agd N06e Labor Year Work A;ailabl; Dispute” Other-— 1969 Illinois 509 3.1 11,407 68.6 40 0.2 455 2.7 U.S. 11,820 6.9 32,686 19.0 4,361 2.5 23,977 13.9 Illinois 622 3.0 13,819 66.4 0 --- 359 1.7 1970 U.S. 10,570 4.9 35,558 16.6 6,168 2.6 31,092 14.5 Illinois 669 2.5 17,067 64.0 70 0.3 625 2.3 1971 U.S. 11,902 4.5 37,261 14.2 4,706 1.8 41,291 15.8 Illinois 727 2.8 15,610 60.8 0 --- 949 3.7 1972 U.S. 12,607 4.4 42,026 14.8 4,911 1.6 46,485 16.4 Illinois 777 2.9 15,602 58.8 0 --- 1,077 4.1 1973 U.S. 15,411 5.0 48,953 15.9 5,990 1.8 47,101 15.3 197“ Illinois 680 2.1 20,839 63.2 0 --- 1,111 3.4 U.S. 14,278 4.0 50,437 14.1 6,236 1.6 54,959 15.3 Data source: Unemployment Insurance Statistics [30]. 58 TABLE 7 Appeals Based on Not Able and Not Available, Illinois vs. Next Highest State Year State Total Not Able and Decisions Not Available # %1_ Illinois 16,629 11,407 68.6 1969 New Hampshire 942 351 38.3 Illinois 20,809 13,819 66.4 1970 Connecticut 5,974 1.974 33.0 Illinois 26,656 17,067 64.0 1971 New Hampshire 1,372 482 35.1 Illinois 25,609 15.610 '60.8 1972 Connecticut 9,422 3,955 42.0 Illinois 26,524 15,602 58,8 1973 Connecticut 8,488 3,499 41.2 Illinois 32,950 20,839 63.2 1974 Connecticut 7,439 2,934 39,4 Data source: Unemployment Insurance Statistics [30]. 59 The model as constructed seems particularly adapted to estimation by the Zellner seemingly unrelated regressions.36 With some variables remaining constant across years for each state and others varying annually the possibility of cross-equa- tion (with each year a separate equation) correlation seems significant. A system of seemingly unrelated regression equa- tions is equivalent to a single equation model if all the regression coefficients in each equation are the same as the regression coefficients in any other equation. In such a case the single equation can be estimated by the method of ordinary least squares on the pooled cross section and time series data.37 Chow38 tests where run on the various equation forms to see if the estimated coefficients are significantly different from year to year. Table 8 shows the results of those tests. That the coefficients do not differ makes the Zellner seemingly unrelated regressions nearly equivalent to ordinary least squares. As a result the equations were estimated on the pooled cross-section time series data set using ordinary least equares. Given the high probability of autocorrelated disturbances it is not surprising that the estimated equations on the pooled 36Kmenta L14] discusses the implications of Zellner seemingly unrelated regressions in some detail. 37See Kmenta [14], p. 518. 38Kmenta [14], pp. 373-374, for a derivation of the simple Chow test. 60 TABLE 8 Chow Tests 1974 (SSE - 2 SSE1)/5K 5K z . p i-1969 is distributed as FXN _ 6K I974 *1974 i z SSEi 2 Ni - 6K i=l969 i=1969 Equation Form Z 1 0.7506 2 0.7373 3 0.7131 4 0.6918 5 0.6636 6 0.6796 7 0.6594 61 data set exhibit quite low Durbin-Watson statistics, on the order of 0.8. In order to reduce the autocorrelation a Cochrane- Orcutt39 type transformation was performed on the data. The actual data used in the OLS estimation procedure are of the form Yi, t ‘ pY1, t-l = 8(Xi, t" 9X1, t-l) with p = 0.7. The transformation removed most of the correla- tion, the Durbin-Watson statistic rising to about 1.9, but it also reduced the sample size from 300 to 250.‘40 Several indicators of experience rating strength are used as shown in Table 9. The average minimum rate AL, the average maximum rate AH, and PMAX, a proxy for ineffectively charged benefits, are as in Chapter 2. TAX/TOT is the ratio of taxable to total wages for each year. BASE is the taxable wage base for each year. Although neither the rates nor PMAX are from the time span of the study, they should indicate the structure of experience rating in each state, assuming that the time period 1969 through 1974 was not out of the ordinary. The average minimum tax rate AanointS out the effective- ness of the lower part of the tax scale and also hints at the 39Kmenta [14] outlines the Cochrane-Orcutt iterative method, uoThe method was not truly Cochrane-Orcutt in the sense that an iterative technique was not employed and a p was not cal- culated from the residuals. Because of computer statistical package limitations an initial 3 was chosen by h e (2 - DW)/2, where DW is the Durbin-Watson statistic from the untransformed regression. Given that starting point, p was varied until the Durbin-Watson statistic indicated little autocorrelation. 62 effectiveness at the upper end. Deviations of the minimum rate from zero imply a range where some amount of improvement or deterioration in a firm's experience record does not result in an altered tax rate. Firms operating at a point on the tax formula where changes in their employment records do not alter their tax rates have no real economic incentive to challenge a benefit claim by a separated worker. The higher the minimum tax rate, i.e., the wider the range where no changes in rates occur, the fewer the employer challenges. Also ineffectively charged benefits must be made by other firms in the UI system. If those charges are spread over firms to include those with perfect employment records, then deviations of the minimum rate from zero hint at the size of ineffectively charged bene- fits. A firm paying at the ceiling tax rate need not worry about further worsening of its employment record since another separation has no effect on its tax rate; therefore the firm has no economic incentive to challenge a UI claim. Thus the larger the deviation of the minimum rate from zero the fewer the benefit award appeals by employers: (1) improved records do not gain lower taxes because of weak rating at the lower end of the tax scale; (2) worsened records do not produce higher taxes because of weak rating at the upper end of the tax scale. PMAX and AH are indicators of the strength of experience rating for firms at the upper reaches of the tax scale. PMAX is included besides the maximum rate itself since the degree of experience rating is pointed out not by the level of the 63 maximum rate, but by its effectiveness. PMAX, as a proxy for the size of ineffectively charged benefits, reflects the effect- iveness of the maximum rate.”1 The expected total benefit payment, EXBEN, is the product of the average weekly payment for total unemployment and the average actual duration of benefits, in weeks, for all bene- ficiaries. 821, 822, and SZ3 are included to indicate the mix of firms according to size in each state. The variables were created from 1972 Census of Manufactures [24] Area Series. SZl is the percentage of firms with 1 to 19 employees, 822 the percentage with 20 to 99, and 82’4 the percentage with 100 to 499 employees. 8Z4, the percentage with 500 or over em- ployees was not included since it is a linear combination of the first three. The state average annual unemployment rate, UER, was included in the control variables to account partly for differ- ences in the mix of unemployment insurance claimants. A possible simultaneity problem exists if the percentage of claims con- tested influences the unemployment rate. The problem is partly relieved by using the survey unemployment rate, rather than the fi" . ’1‘. L ulSee Chapter 2 for a more detailed discussion of all the indicators of experience rating strength. The indicators are entered individually because of the difficulty involved in constructing a valid single index, and also because historically there has been an interest in their separate effects. 64 insured unemployment rate.42 DENR is the number of denials per 1,000 claimant contacts. Since the employment security commis- sion itself first passes on the eligibility of a claim, close scrutiny and strictness on their part should preclude many employer challenges of benefit awards. PRB is a proxy for what the employer perceives as the probability that the challenge of an award will result in a reversal. PRBi, t was created as PRBi’ t ' %Ik§1 PRBLi, t-k’ where PRBLi, t is the percentage of employer lower authority appeals which were ruled in the employer's favor. It was included under the assumption that employers are influenced by what they see as their chances of actually winning a contested claim case. The average wage in covered employment WG was used to control for wage differentials, especially across states. Since the UI tax is levied on wages only up to a certain amount, a $4,200 wage base in one state is not really equivalent to a $4,200 wage base in another state if their wage structures differ. Regression Results The results of regressions on the various equation forms .appear in Table 9. The figures in parentheses below the estimated quee Holen and Horowitz [12] for a discussion of changes in claimant mix. They are concerned about the effect that the un- employment rate has on the mix and the consequent effect that the mix has on denials. High unemployment should reduce denials, they concluded, because of the superior quality of the average claimant. A similar effect should result with regard to em- ployer challenges of benefit awards. Successful challenges how- ever, should reduce the covered unemployment rate. In that vein, Holen and Horwitz did find evidence that high denial rates re- duce the unemployment rate. 65 Am=.oav Ams.ouv “so.o-0 “mo.ouv Aa=.oa0 Amo.on0 Amo.oav mmsm.oa Hmmm.oa .omso.ou ammo.ou Hamm.oa memo.on Homo.ou =< Amm.ma0 Amm.mu0 mam.euv Amm.s:0 mam.muv Ame.suv amm.=av mmmo.su smoo.sn ommm.mu esmm.mu msmo.=u mamm.mn mmmm.mu ma Aoa.o|0 Aao.o0 Amm.o0 Amm.o0 Amm.on0 Amm.o0 AHm.o0 mmmo.ou amao.on ammo.o mmmo.o momm.ou mmmo.o mamo.o mesa Asm.sv Amm.s0 Amo.30 Amo.=v Ams.e0 AHH.=0 hos.=0 Hmoo.o omoo.o omoo.o memo.o Hmoo.o omoo.o omoo.o zmmxm Amm.mnv Ham.mnv Aa:.muv Hom.mnv amm.mn0 Amm.mnv Amm.mnv maam.o ommm.ou aaom.ou mmam.on mmam.ou omam.on mmmm.ou mm: Ama.ov Asm.ov Amm.ov Amm.ov Aom.ov Amm.o0 Amm.ov mmso.o memo.o mmso.o =aso.o meso.o maso.o mmmo.o mma Ham.m0 Amm.m0 Amm.m0 1am.m0 1mm.m0 Amm.mv Aom.m0 ammo.o ammo.o mmmo.o ommo.o ammo.o mmmo.o ammo.o mzmo Aa=.H0 Asa.mv Amm.s0 Amm.mv Amm.sv mom.s0 Asm.sv oomm.s mmms.m mmms.s amom.s Hams.s masm.H mmam.m mam Amo.sv AHN.HV Amm.av Aem.s0 Ams.mv mom.mv Amm.sv mama.o mamm.o mmmm.o momm.o mmmm.o Hmmm.o mmom.o mam Asa.mv Amm.mv Amm.mv Aeo.m0 mam.s0 Amm.H0 Amo.m0 HoaH.H msam.H 0:0N.H masm.H smmm.H «wom.H mHmm.H Ham Aav 100 Amv i=0 Amy Amv Amy mandates muHSmom :oHnmoswom m mqm<8 .m someone as no mm .mm .meza 66 Hmmu mmem mmmu mnemsOEoE HmuoosoHansn 0:0 0al0mmH mumo HmHocmcmn HD mo xoonccmm BOB\X HoosoHpcoov m mHm:H.0 m0 Amm.sv o mmom.o 00 Asm.o0 o . ammo.o 0:0 A00.010 o: memo.ou + h0>0.0l A< mm:H.0 + 0000.0! 0:0 AHm.0|0 0| 0H0m.07 :70 Amm.suv m: mmma.mu muv A00.m|v ml 00H0.:: + :000.:l x.0 hESEHCHEv CO . O a me on cooasms>m A 0, i.e., workers may get utility or disutility from beifié unem- ployed. The only productive resource is labor, e, so that the production functions are X I Fl(el) Y I F2(e2) We wish to maximize consumer A's utility 83 (i) UA I UA(xA, yA, lA’ nA), subject to consumer B's utility, (11) U5 I UB(xB, YB, 18, nB); the production functions, (iii) X I Fl(el); and (1V) Y a F2(€2); the resource constraints, | >< (v - ix) xA + xB - and the unemployment-time transmission functions; (x - xiii) n I G 11 11(X) n21 921(X) H22 3 H22(Y) n12 = H12(Y) where :11.1 I unemployment transmitted to firm i from firm j; the additional resource constraint, 84 (xiv) n11 + n21 + n22 + n12 I N and the total resource constraint, (xv) E + N + L - R. Equations x - xiii indicate that a firm creates unemploy- ment through its own production process, and that the firm itself may not suffer all the unemployment created. Thus Firm 1 causes n11 + n21 units of unemployment but only incurs n11 of the units itself. The other n21 units of unemployment created are forced onto Firm 2. Transmitted unemployment may take any of several forms. For example, a strike in Firm 1 may result in induced layoffs for Firm 2. Layoffs in one firm due to technological reorganization (e.g., changeover in the auto industry) might partially cause layoffs in some related industry. We assume that none of the unemployment tax can be shifted forward onto consumers or backward onto labor. We also assume that the price of unemployment (the UI benefit) is either determined in the market, or the government sets the price at what would be the market price. The solution to this set of 15 equations shows that con- sumer A should set (1) aUA . aUA aU —- F' - __ axA Ta A 1 + Ta A 3% all + Gél 85 (2) 00,, aU ' a A 3U 3U r— T F + A - A yA 8 A é BIA anA H52 + H12 Thus consumers set marginal utility of a good equal to the marginal production cost of the good plus the marginal cost of the unemployment created by that good's production. 80A 3UA , the difference between the marginal utility of FIX! anA leisure and the marginal utility of unemployment time represents the price of unemployment for the consumer. If A is indiffer- ent between leisure time and unemployment time, he only pays attention to the production cost of the good in deciding his optimal purchases. The greater the marginal utility of un- employment time, the smaller is the second term in (l) and (2), and hence the more he desires of goods X and Y, since he cares less about any unemployment time their production forces upon him. The smaller the marginal utility of unemployment time (and it could be negative), the greater the right hand side in (l) and (2) and the less he desires of each good be- cause of the forced leisure inherent in each production process. The individual unemployment transmission functions, 011, 521’ 322, and H12 are not observable by the taxing authority. Instead, what it Observes is the unemployment suffered by each firm, G11 + H12 for Firm 1, and H22 + 021 for Firm 2. A per- fectly experience rated UI tax charges each firm with the cost Of the unemployment that it incurs, which is not necessarily the cost of the unemployment that it creates. 86 In a regime of perfect experience rating Firm 1 faces the profit function (3) 71 = pxX - peel - pnCGll(X) + H12(Y)J where px is the price of output X, pe is the wage rate and pn is the price of a unit of unemployment (the UI benefit). Similarly Firm 2 faces (4) «2 = pyY — peez - anH22 0. Socialized unemployment is an increasing function of the size of the labor market. 55As quoted in Nelson [13], p. 3. 91 To focus on the effects of socialized unemployment costs, assume that unemployment created by each individual firm is ~tranSmitted to itself only. Transmitted unemployment is pro- bably small and its effect is further diminished by the counterbalance of received unemployment. Secondly, unemploy- ment which a firm inherits due to the instability of a supplying industry or demanding industry can be attributed to the firm's market and hence be a responsibility of the firm for dealing in that market. Thus the costs of seasonal unemploy- ment in the canning industry are rightly attributed to the canning industry, even though the seasonality is a result of seasonality in agriculture. Both reasons make the assumption 012 I H21 I 0 not unreasonable. To incorporate socialized unemployment costs into the model and to elimdnate transmitted unemployment, we replace equations x - xv with the unemployment-time functions, (x-a) nl I G(X) (xi-a) n2 I H(Y) (xii-a) S . S(E) the additional resource constraint, (xiii-a) n1 + n2 3 N and the total resource constraint, 92 (xiv-a) E + N + L = R The solution to this new set of 14 equations shows that consumer A should set (7) aUA aUA ' '3U au 0' + as ' 3xA 31A // l laIA. anA . ae1 //F1] (8) aUA aUA . [an au ][ as '1 3VA aIA //F2 + 311 ShA ae1 //F2 Optimality requires that consumers set the marginal utility of a good equal to its marginal production cost plus its marginal cost of unemployment resulting from that good's production. ES F. indicates that consumers weight the marginal cost of 381 -..-‘.c the larger labor market by the marginal product of the new unit of labor, since the new unit imposes a cost because of the now larger labor market, but also provides additional output for consumers to enjoy. The larger the unit's marginal product, say F1, the smaller is aUA/axA, indicating the consumer is more willing to accept the increased labor market maintenance expense in exchange for more X. S(E) represents the unemployment attributable to the maintenance of the labor market itself. aS/ae represents the marginal socialized unemployment created from the use of labor unit e, and says nothing about which firm actually suffers that unemployment. The socialized unemployment that results from the use of the labor market is not necessarily distributed across firms according to their participation in the market. And if it is not so distributed one firm suffers more socialized 93 unemployment than it creates by its use of the market. As in the above example, the construction industry might incur a disprOportionate share of the socialized unemployment resulting from a tightened money supply. Let socialized unemployment S be distributed across firms according to functions 131(3), 1 a l, 2 Since consumers care only about who creates unemployment, the H functions do not alter their optimal choices. When the insurance taxing authority uses a perfectly experience rated unemployment system, he cannot observe G(X), H(Y), S(E), or D1(S) and D2(S). [Instead he taxes each firm according to the unemployment it suffers, whether it is re- sponsible for it or not. Perfect rating forces the firms to face profit functions (9) «1 = pxx - peel - p,EG)1. Producers maximize their profits when a ' 1 I I (ll) px pe/Fl + anG + D1 3: Fl] 1 g t + t + D" as F! (12) py pe/F2 anH 2 11, deg which implies that the firm produces the optimal amount of goods only if Di BS/ael I aS/ael and Dé aS/ae2 = aS/aez, if each firm suffers only the socialized unemployment it creates 94 through its participation in the labor market. It is not the size, relative or absolute, of the socialized unemployment that dooms a perfectly experience rated 01 tax as an ideal cost allocator, but it is how that unemployment is distributed across firms. Even if all unemployment were socialized unemployment, a perfectly experience rated tax would still lead to an Optimal allocation of unemployment costs as long as socialized unem- ployment is distributed to each firm in accordance with how much socialized unemployment the firm is responsible for by its participation in the labor market. Thus if G I H I 0, perfect rating would still lead to an optimal resource alloca- tion as long as Di BS/ael I as/ael and Di aS/ae2 I aS/aez, Let us assume that unemployment S resulting from labor market maintenance is a linear function of the size of the labor market. S(E) I S(el, e2) I k(e1 + e2), which says that firms create socialized unemployment proportionate to their use Of the total labor market. Really this assumption is the ra- tionale for instituting a flat rate UI tax. If socialized un- employment is the bulk of unemployment, and if socialized un- employment is created in said proportion, then a flat tax will, in general, more accurately charge each firm's unemployment costs. Suppose the taxing authority institutes a flat rate tax on employers to pay the costs of unemployment insurance. Each firm pays a fraction Of the total unemployment costs equal to its fraction Of the labor force. The firms face the profit functions (13) «1 = pxX - peel - pn[(G(X) + H(Y) + S(E)) el/E] 95 (14) w I p Y - pee 2 y - pn[(G(X) + H(Y) + X(E)) eZ/EJ. 2 Producers maximize profits when (15) px I pe/Fi + pnCG' + ggl Fijel/E + anG+H+SJe2 E2F’; (16) p I p /F' + p [H' + as F']e /E + p [G+H+s]e y e 1 n 3g— 2 2 n l 3235 The optimality conditions (7 - 8) indicate that producera} marginal rate of transformation should be (17) F' I (py - an') (pe + pn aS/ael) (px - pnGT) (pa + pn aS/ae2)' H 54 Given the assumption S I k(el + e2), (17) becomes (17a) F1 I 31‘ an' F; p x ' pnG1 Given a flat rate tax, the profit maximizing conditions in this case (15 - 16) indicate that producers hire labor so that together: (18) Fi a (py - anv :3) (9e + pugs . el + (G+H+S)el) E 61 E‘ E? - ' + + + + F2 (px pnG ‘:l) (pe pngg . e2 (G'H SR27 E 2 ‘E‘ E? If the fraction of unemployment [G + S(el)]/[G + H + 8] caused by Firm 1 is equal to its share of the labor force el/E, and 96 similary for Firm 2, a flat tax will lead to an optimal re- source allocation. Equations (11 - 12) show that in a regime of perfect exper- ience rating producers set the ratio Of marginal products so that (19) Fi . (3y - an') (pe + aniaS/ael) 173’ (px - pnG'5 (De + anZIaS/Sez) Since S(E) I k(e1 + e2), by assumption, the flat tax should be a better cost allocator than the experience rated tax when socialized unemployment S(el) dominates own unemployment 0. Conversely, (19) shows that an experience rated tax is appro- priate when socialized unemployment is relatively small. For the polar cases, when S I 0, i.e., there is no socialized un- employment, the experience rated tax always leads to the optimal resource allocation. When 0 I H I 0, i.e., there is no own firm created unemployment, the flat tax rate leads to an Optimal allocation, given the assumtions S I k(el + e2), since the second term in (18) then equals 1. If the socialized unemployment is communicated to firms in proportion to their fraction of the labor force, i.e., DiaS/ael aS/ael and DéaS/aez I aS/ae2, then an experience rated tax also leads to the optimal cost allocation. As mentioned earlier the assumption S I k(e1 + e2) is precisely the rationale for a flat tax. If socialized unemployment arises according to some other mechanism, the conclusions about the allocative effects of the flat rate and the experience rated tax schemes may differ. 97 For example, suppose socialized unemployment is created in proportion to own caused unemployment, i.e., S I k(G(X) + H(Y))- In this case the choice of tax plan depends crucially on D1 and D2. The distribution of socialized unemployment implied by D1 and D2 may so poorly reflect as/aG(X) and aS/aH(Y), that the flat rate tax better allocates the costs of unemployment.56 Combination Tax Schemes If both own caused unemployment, G and H, and socialized unemployment S(E) are significant, the choice as to which tax scheme works better is not at all clear. Since a flat rate tax does a good job of allocating socialized unemployment costs, and an experience rated tax does a good job of allocating own caused unemployment, a mixture of the two taxes would seem the best solution of allocating both kinds of costs. In fact, from the very inception of the U.S. unemployment insurance system the idea of a mixed flat rate and experience rated tax was considered. President Roosevelt's Committee on Economic Security, when investigating how to establish a national system 56Suppose G(X) + kG(X) > H(Y) + kH(Y), but G(X) + 01(3) < H(Y) + 02(3). Now the flat rate tax may not do a very good job of allocating unemployment costs, but it beats the experience rated tax on the averages. 98 of unemployment insurance, recommended a minimum 1% contribu- tion rate by each employer into a pooled fund.57 Given both significant social unemployment costs and own caused unemployment costs, the question of how to levy the unemployment tax becomes one of how much experience rating is apprOpriate. Suppose the total amount of social unemploy- ment costs can be observed. Then setting a flat rate tax to insure that each firm pays its share is a simple matter of. computing the cost per labor input. Experience rating the remaining unemployment does not necessarily achieve optimality. Given our simplifying assumptions, a firm's charged unemploy- ment is the sum of its own caused unemployment plus the socialized unemployment that is communicated to it. Thus the firm profit functions under the mixed tax scheme look like «x pxX - peel - anGCX) + D1(S(E)) - els(E)/E] a ll y pyY - pee2 - anH(Y) + 02(s(E)) - e2S(E)/E] Optimality is achieved only if social unemployment is distri- buted across firms in proportion to their use of the labor force. But from above, if that condition is met, a perfectly experience rated tax alone also leads to an Optimal output mix. To see why this condition is required for the mixed structure to achieve optimality, suppose firm A creates 3 units of unemployment per labor input and firm B creates 7 units. There are 3 units of socialized unemployment per labor input in the economy. Now suppose socialized unemployment is S7Altmeyer [2], p. 24. 99 distributed unevenly such that firm A inherits 4 units per input and firm B inherits only 2. Firm A's charged unemploy- ment is 7 and firm B's is 9. A flat rate tax of 3 units per input insures that each firm covers its socialized unemployment costs. But by experience rating the remainder firm A in effect pays for causing a unit of unemployment which it in- advertently inherited because of an inequitable distribution of socialized unemployment. Firm A pays a tax of 7 when it Should be 6, and firm B pays 9 when it should pay 10. To gener- alize, if every firm has charged unemployment at least as great as its socialized unemployment responsibility then a perfectly experience rated tax and a combination tax both lead to the same result. For a mixed tax scheme to approach more closely the proper allocation of unemployment costs than a straight ex- perience rated tax, at least one firm must have charged unemployment less than the socialized unemployment for which it is responsible. Since there are 13 states with minimum tax rates of 0, then that situation definitely can exist. Suppose the taxing authority sets a flat tax rate so that each firm pays the exact amount of its socialized unemployment costs. Any unemployment costs over the minimum are assumed to be own caused and are experience rated. The result is that firms that previously paid no tax now pay the correct amount, firms that previously paid less than socialized unemploy- ment costs now pay closer to the correct amount, and firms that previously paid more than the flat rate now pay the same amount. 100 Unfortunately this gives the taxing authority a surplus of tax collections over unemployment costs. The mixed tax scheme has clearly improved the unemploy- ment cost allocation, since now at least all employers are covering their socialized costs. Just as clearly, the prOper allocation has not been achieved since the taxing authority cannot distinguish between caused unemployment and inherited social unemployment, but still only observes charged unemploy- ment. He would like to pass out the surplus as credits to those firms that inherit more than their equitable share of social unemployment. But even if he knew what those firms were, he most likely would not have a large enough surplus to pass out. For suppose the flat rate is 3% and some firm has charged unemployment requiring a 4% tax rate. This firm creates unemployment such that a tax rate of 2% would cover the cost. That cost added to its socialized cost says that the firm should pay a 5% tax, not a 4% tax. In effect the firm is escaping 1% of its socialized costs because it inherits less than an equitable amount of those costs. The Optimal Amount pf Experience Rating Given that only charged unemployment can be observed, the optimal amount of experience rating is a two part problem. First, how high should the minimum tax rate be, which is actually the question how much unemployment should be considered social unemployment? Secondly, how should the surplus tax collections be distributed? (0r really, how Should taxes be collected so that no surplus results?) Lowering all tax rates so as to 101 collect no surplus assumes that social unemployment is distri- buted fairly evenly among all firms, since all are given a tax reduction, except those at the minimum. Tax rate ceilings, i.e., limiting taxes for firms with a large amount of charged unemployment, assume that the reason those firms have so much charged unemployment is because they suffer a grossly dispro- portionate amount of social unemployment.58 The foregoing analysis of experience rating and uniform tax rates as cost allocators in the presence of social unemploy- ment assumed that the unemployment generating functions G(X) and H(Y) were fixed. Chapter 2, however, shows that this is not necessarily the case. The evidence from that study in- dicates that the costs imposed by the experience rated UI tax may induce a firm to stabilize its seasonal employment fluctua- tions, i.e., alter its unemployment generating function, in an effort to reduce its UI tax burden. The gain from an experience rated UI tax then might not simply be a better allocation of unemployment costs, moving closer to the optimal point along a transformation curve, but moving from an inefficient point inside the curve to a point along the efficient frontier. If Firm 1 alters its unemploy- ment generating function G(X) so that the new Gl(X) < G(X), the time that previously was wasted in unemployment can be used 58If cyclical unemployment is socialized unemployment then ceiling rates protect cyclically sensitive industries from bearing the brunt of a general economic downturn. If extremely seasonal unemployment is partly socialized unemployment as Haber and Murray [9], p. 353 seem to argue, then ceiling rates pro- tect those extremely seasonal firms. 102 produce more consumer goods X and Y, without either consumer relinquishing any leisure time.59 Therefore the marginal rates of transformation resulting from an experience rated tax and some alternate tax scheme may not be comparable. Figure 1 illustrates this point. Let T be the efficient frontier and let A be the optimal point (from consumers' standpoint) along T. Suppose the marginal rate of transformation of X for Y at inefficient point B is equal to the MRT at point A. Suppose point C along the efficient frontier results from allocation UI costs through experience rating. Point C is clearly super- ior to point B, yet on the basis of MRT's B appears superior. FIGURE 1 59This assumes that the marginal utility of leisure time is greater than the marginal utility of unemployment time. If the opposite is true then consumers prefer more unemployment time and less leisure. In this situation an experience rated tax can also lead to more efficient production. If costs are not allocated prOperly, then some firms may 'over stabilize' employment, i.e., cut employment fluctuations to a greater degree than consumers wish. 103 Even if experience rating 'misallocates' resources (in the MRT deviation sense) worse than some alternate tax plan, it may lead to a superior output combination. (See Figure 2.) Suppose P; and p; are the prices of goods X and Y at the optimal point C. Then a line V with slope -P§/P; is tangent to T at C. Let V' be parallel to V and run through B, the inefficient point. Y' represents the value of output combination B evaluated at the optimal prices. All combinations to the right of V' are of greater value, in terms of the optimal prices, than B. Thus as long as the experience rated tax moves firms to produce somewhere on the efficient frontier between t1 and t2 (or for that matter anywhere to the right of V'), the value of that output is greater than the inefficient point combination even if the resource allocation appears worse. FIGURE 2 Conclusions An experience rated unemployment insurance tax unambigously leads to an optimal allocation of unemployment costs only if 10“ each firm is charged with the unemployment which it creates. In the presence of socialized unemployment, when the unemploy- ment is proportional to the size of the labor market, a mixed tax scheme of a uniform tax to cover the social unemployment cost and an experience rated tax to cover the rest, is likely to lead to a better allocation of cost than either plan by itself. Thus financing the administrative costs of the state UI programs through the Federal UI tax (which is effectively a 0.7% uniform tax), and the benefit costs portion of the pro- grams through an experience rated tax is appropriate. If the cost of spells of unemployment greater than 26 weeks are assumed to be socialized unemployment costs due to cyclical fluctua- tions in the economy, then federally funded extended benefit programs are appropriate. If those unemployment costs are not socialized costs, then extended benefits should be financed through experience rated state taxes. Also, if socialized unemployment costs are costs that should be borne by all the participants in the labor market, then contributions are called for from both employers and workers. Finally, since experience rating can induce firms to stabilize their own employment fluctuations, it can shift the equilibrium output combination to a combination on or nearer the efficient frontier. Looking at the allocation of U1 costs as represented by marginal rates of transformation without considering the real gain from the experience rated tax can be misleading, since a suboptimal allocation might be superior to what appears to be an optimal point. CHAPTER V Conclusions The 1932 Wisconsin Unemployment Compensation Act called for an experience rated UI tax, one which ties an employer's tax rate to his past employment experience. Since that original UI statute, experience rating has gained favor primarily on the basis of three arguments. The Congress of the United States encouraged (actually all but forced)60 the states to experience rate their unemployment insurance taxes under the assumption that an experience rated tax would impel each firm to stabilize its own employment fluctuations. The 'Wisconsin School'61 originators of the idea of an experience rated tax emphasized that it could allocate the costs of unemployment to those firms which caused the unemployment. Finally, an experience rated tax encourages firms to take a direct interest and to partici- pate in the administration and development of the UI system in its own state. 60See Haber and Murray [9], p. 338 for the Senate Finance COmmittee's statement of purpose in favoring experience rating. Becker [H], p. 12-16 outlines the debate over allowing federal UI tax credits only through experience rating. 1 Especially Arthur Altmeyer, John Commons, Harold Groves, Paul Raushenbush and Edwin Witte. 105 106 All these pro-experience rating arguments imply that firm behavior is somehow altered by an experience rated unemployment insurance tax. Yet no microeconomic study of the effects of experience rating on firm behavior has been done. The purpose of these three essays was not a cost benefit analysis of ex- perience rating, nor even an evaluation of each of the arguments. The purpose was to look at some claimed implication of each argument to see if those implications are valid. The first two essays asked whether the experience rated UI tax is significant enough to affect firm behavior in stabiliz- ing employment and in policing benefit claims. Both are empirical questions and were handled in an empirical format. The parameters of the experience rated UI system, that is the tax rates and the taxable wage base, were used to test the effects of rating. The third essay used a general equilibrium analysis to ask how experience rating might misallocate the costs of unemployment if layoff records do not reflect the un- employment costs generated by a firm. Review of Findings An experience rated UI tax varies each firm's tax rate so that its payments into the UI fund balance its expected benefit withdrawals. The profit maximizing firm attempts to minimize this tax burden along with its other costs, and in so doing stabilizes its employment. If employment stabiliza- tion can be realized at all, it should certainly show up in damped seasonal employment fluctuations. Here there are more 107 possible ways to smooth seasonal swings than other types, like cyclical, for instance, because they are more predictable and their source more easily pinpointed. A firm might try to alter its production schedule to reduce peak period demand for labor, or it might try to take on complementary lines so that when one output is in season the other is out and vice versa. The first essay showed that seasonality of employment is affected by the experience rated UI tax. Three industries were studied: SIC 161, highway and street construction, ex- cept elevated highways; SIC 233, women's, misses', and Juniors' outerwear; and SIC 2&3, millwork, plywood, veneer, and other prefabricated wood products. Only two of the three, SIC 161 and SIC 233, showed any reaction to experience rating, and in each the reaction was different. Since each industry faces different limitations on its ability to stabilize unemployment, i.e., different stabilization costs, it is not surprising that each exhibits a different response to the tax. The wood pro- ducts industry, SIC 2h3, which showed no reaction to experience rating, faces seasonality in its own production and in the demand for its output. Apparently for this industry the costs of stabilization are too high no matter how effective the ex- perience rating is in charging firms for their unemployment costs. On the other hand, strong rating does increase employer stabilization efforts in the construction industry, SIC 161, despite its strong weather induced seasonality. The most definite response to experience rating was in the outerwear industry, SIC 233. Firms in this industry were represented 108 in R6 states in 197“ even though those states exhibited a wide variation in experience rating strength. Ostensibly, firms in states with strong experience rating have survived by stabilizing their own employment and reducing their UI tax burden. The second study showed that experience rating increases employers' propensity to challenge benefit awards.62 If a firm successfully challenges a claim by an ex-worker, no bene- fits are charged to its account. In states with strong exper- ience rating the firm pays a larger portion of the cost of its separated workers' UI benefits, and hence has a greater incentive to examine claims for violations of eligibility standards. Experience rating has come under bitter attack on this point63 by some, yet others have praised this policing incentive.6u The potential for abuse of the appeals process exists, but the data show that very few benefit awards are challenged by em- ployers.65 If all states widened their tax rate spreads to the 0.3% to 5.h% that Michigan averaged in the study, appeals would increase by only 6,000 to 11,000 annually. 62This result is evidence that firms believe that shifting of the UI tax is not complete and immediate. 63Recall the statements by Walter Reuther and Andrew Biemiller as quoted in Chapter 3. 6“For example, see Bailenson [3]. p. 5. 65In 197“ there were 17.3 million new spells of insured unemployment. 3.1 million of those claims were disqualified by state authorities. Just 75,566 employer appeals were decided that year, only 0.5% of all benefit awards. Source: Unemploy- ment Insurance Statistics [30]. 109 The final essay considered the merits of experience rating as an allocator of unemployment insurance costs. Pro- ponents of rating claim that tax rate ceilings and tax rate floors create interfirm subsidies, forcing stable firms to partially pay the unemployment costs of unstable firms. As a result, products from stable industries are priced artificially high and resources are misallocated away from stable indus- tries and into unstable industries. Since labor is among those resources, incomplete experience rating increases un- employment. The experience rated tax allocates UI costs to firms according to their layoff records. If layoffs do not accurately reflect unemployment caused by the firm then those costs will not be properly charged. This situation arises especially if there is socialized unemployment, unemployment not attributable to any one firm using the labor market, e.g., cyclical unem- ployment. If the costs of socialized unemployment are pro- portional to the firm's use of the labor market as indicated by its payroll, then a flat rate tax allocates those costs better than an experience rated tax. If there is both socialized unemployment and unemployment which can be attributed to firms, then a mixed tax plan is best, with a flat rate to cover the socialized costs and an experience rated tax to cover the rest. llO ' RecommendationS‘for Change These three studies have demonstrated the merits of an experience rated UI tax in accurately allocating unemployment costs, in stabilizing employment, and in the fostering of employer interest and participation in the UI system. They indicate that the real question about experience rating should not be whether it has any effects, but whether it has enough of an effect to outweigh its additional administrative costs.66 At present data on the parameters of the system are simply too crude to allow reasonably accurate estimates of the size of the effects of rating on firm employment stabilization and on the propensity to appeal benefit awards. The first recom- mendation addresses itself to this problem. 1. More extensive data on the parameters of the exper- ience rated UI system are needed. Data which could be fairly easily compiled from the states include the size of negative balances, to indicate the effectiveness of the maximum tax rates, and the size of noncharged benefits. The distribution of firms by tax rates would be helpful in indicating how sensitive the tax is to changes in employer experience. A wide range of tax rates can be deceiving if most of the firms are concentrated at one or a few rates. Since experience rating is designed 66Becker [A], p. 323 estimates that about HZ of the total administrative costs of the UI system are due to experience rating. He cites that several state administrators claim that computerization of UI systems is reducing the additional administrative load. 111 to affect firm behavior, more disaggregated tax data would help to estimate its effects. Minimum and maximum tax rates by industry, and the variation of rates within an industry would be helpful, though clearly more expensive to compile than the aggregate state data suggested above. 2. Experience rating should be strengthened. The federal government should require states to reduce the minimum rate to zero and eliminate ceiling tax rates. The taxable wage base should be raised. A strongly experience rated tax can reduce unemployment in two ways: by inducing individual firms to stabilize their own employment fluctuations, and by eliminating the interfirm subsidies that are the result of ceiling tax rates. These subsidies allocate labor resources away from stable employment industries and into unstable industries. These employment gains can be realized without greatly increasing employer challenges of benefit awards. 3. Socialized unemployment costs should not be subject to experience rating, but should be financed through a flat rate tax. Although other observers of the UI system might disagree, we would define socialized unemployment costs as noncharged benefits and benefits paid after 26 weeks of un- employment. The noncharged benefits should be financed at the state level to prevent states from capriciously noncharging benefits in hopes of being subsidized by other states. Bene- fits after 26 weeks should be financed completely by the federal flat rate tax in order to protect states with con- centrations of cyclically sensitive industries. 112 In the very success that unemployment insurance has en- Joyed over the last forty years as a "first line of defense" against temporary Job loss lies a danger. In recent years UI has become a convenient frame on which to hang additional in- come maintenance programs.67 Since all the benefits of ex- perience rating depend on its ability to accurately allocate unemployment costs, the additional socialized unemployment costs should not be incorporated in the regular state UI sys- tems. In 1935 President Franklin Roosevelt requested from the Congress an unemployment compensation law with a goal "to af- ford every practicable aid and incentive toward the larger purpose of employment stabilization."68 Congress responded with a federal-state plan that emphasized experience rated state taxes to finance benefits. It appears that the ex- perience rated UI tax as presently constituted has at least partly fulfilled the request, and that strengthing of the rating provisions can affect even more stabilization. 6ZBecker [A], pp. 321-323; Hamermesh [10], pp. 106-108; and Warden [21], pp. 95-96, all note the danger of allowing the UI system to take on aspects of a welfare program. 68 See Nelson [16], p. 212. APPENDIX 113 APPENDIX Spectral Measure 2; Seasonality Industry 161 Industry 233 Industry 2H3 New England Maine New Hampshire Vermont Massachusetts Rhode Island Connecticut Middle Atlantic New York New Jersey Pennsylvania East North Central Ohio Indiana Illinois Michigan Wisconsin 'West North Central Minnesota Iowa .337556 .3399““ .338509 .358363 .338922 .3235“3 .3q7985 .353371 .3U6093 .3u1233 .3u503“ .326864 .33329“ .353259 .3U77l9 .335421 .22U126 .191251 .222335 .220586 .290159 .17H067 .222217. .256192 .287136 .2sou91 .188510 .183579 .19u603 .220586 .2027u1 .15ul70 .204261 .20H077 .205291 .2u6605 .178352 .215330 .238035 .282971 .19u16u .188701 .193786 .25772N .210314 11“ APPENDIX (continued) Industry 161 Industry 233 Industry 2&3 Missouri North Dakota South Dakota Nebraska Kansas Pacific Washington Oregon California Alaska Hawaii Sggth Atlantic Delaware Maryland District of Columbia Virginia West Virginia North Carolina South Carolina Georgia Florida .32963h .285158 .330797 .320287 .320229 .31880“ .é61658 .2N5782 .170366 .30H677 .327725 .2uu195 .28h313 .3ou322 .2u8658 .237u79 .289331 .1791u2 .182199 .éo8?50 .18u297 .1857M0 .17510“ .1937u2 .l77l6u .l68098 .175005 .178305 .177993 .201399 .185091 .178702 .2ou103* .225820 115 APPENDIX (continued) Industry 161 Industry 233 Industry 2M3 East South Central Kentucky Tennessee Alabama Mississippi West South Central Arkansas Louisiana Oklahoma Texas Mountain Montana Idaho Wyoming Colorado New Mexico Arizona Utah Nevada .326956 .305292 .283u92 .288835 .318122 .262903 .2H4719 .264099 .320583 .307990 .336656 .287712 .301u16 .184u11 .316585 .281535 .168669 .180229 .166517 .152929 .169997 .170207 .189827 .l6u886 .207670 .212373 .153u1u .189170 .2176?“* .19812H* .19uou2 .182033 .16U809 .l7u9lh .20988l* .17U213 .185065 *Value on basis of 24 month season. Data source: Employment and Wages [28]. BIBLIOGRAPHY [l] [2] 13] \m ‘f5] [61 ~{7J [8] [9] [10] [ll] BIBLIOGRAPHY Adams, Leonard P. Public Attitudes Toward Unemployment Insurance. Kalamazoo,'Michigan: W. E. Upjohn Insti- tute for Employment Research, 1971. Altmeyer, Arthur J. The Formative Years pf-Social Security. Magéson, Wisconsin: ’University of Wisconsin Press, 19 . Bailenson, Stewart L.‘ How to Control the Cost of Unemploy- ment Compensation Claims and Taxes on Your Business. Homewood, Illinois: Dow Jones-Irwin, Inc., 1976. Becker, Joseph M. Experience Rating in Unemployment In- surance. Baltimore, Maryland: Johns Hopkins University Press, 1972. Brechling, Frank. "Unemployment Insurance Taxes and Labor Turnover: Summary of Theoretical Findings." Industrial and Labor Relations Review, July, 1977. Cartwright, Philip W. "Unemployment Compensation and the Allocation of Resources." in M. Abramovitz, The Allocation of EconOmic Resources. Stanford, CaIIfornia: Stanford University Press, 1959. Feldstein, Martin. "The Effect of Unemployment Insur- ance on Temporary Layoff Unemployment." Cambridge, Massachusetts, Harvard Institute oi‘Economic Research Discussion Paper Number 520, 1976. Granger, C. W. F. and M. Hatanaka. Spectral Analysis p§_Economic Time Series. Princeton,’New Jersey: Princeton University Press, 1968. Unemployment Homewood, Illinois: Haber, William and Merrill G. Murray. 'Ipsurance in the American Econq_y. Richard D} Irwin, Inc.,'1966. Hamermesh, Daniel S. Jobless Pay and the Economy. Baltimore, Maryland: Johns Hopkins University Press, 1977. Hamermesh, Daniel S. "Spectral Analysis of the Relation Between Gross Employment Changes and Output Changes, 1958-1966." Review of Economics and Statistics, February, 1969. 116 [12] [13] [1“] [15] C161 [l7] [l8] [19] [20] [21] [22] E23] [2“] [25] 117 Holen, Arlene and Stanley Horowitz. "The Effect of Unemployment Insurance and Eligibility Enforcement on Unemployment." The Journal of Law and Economics, October, 197A. Jenkins, Gwilyn M. and Donald G. Watts. Spectral Anal sis and its Applications. San Franc sco, CaIifornia: _Holden-Day, 1968. Elements of Econometrics. MacmiIlan, 1971. Kmenta, Jan. New York: New York, Lampman, Robert J. (ed.). Social Security Perspectives: Essays by Eggin E. Witte. Madison, Wisconsin: University of Wisconsin Press, 1962. Nelson, Daniel. Unemployment Insurance: The American Ex erience, 1915—1935. Madison, Wisconsin: Univer- sity of Wisconsin Press, 1969. Nerlove, M. "Spectral Analysis of‘Seasonal Adjustment Procedures." Econometrica, July, 196A. Rainwater, P. L. The Rebuttal." "The Fallacy of Experience Rating: Labor Law Journal, October, 1951. Raushenbush, Paul A. tion in Wisconsin." April - May, 1967. "Starting Unemployment Compensa- Unemployment Insurance Review, von Furstenberg, George M. "Stabilization Characteristics of Unemployment Insurance." Industrial and Labor ' Relations Review, April, 1976. Warden, Charles, Jr. "Unemployment Compensation: The Massachusetts Experience." in Otto Eckstein, Studies in the Economics of Income Maintainance. Washington, D.C.: The’Brookings Institution, 1967. Welch, Finis. "What Have We Learned from Empirical Studies of Unemployment Insurance?" Industrial and Labor Relations Review, July, 1977. Michigan Employment Security Commission, 197“ Annual Report. U.S. Bureau of the Census, Census of‘Manufactures, 1972. Washington, D.C.: U.S. Government Printing Office, 1975. U.S. Bureau of the Census, Statistical Abstract of the United States, Washington, D.C.: U.S. Government Printing Office. [26] [27] [28] [29] [30] 118 U.S. Congress, Senate, Committee on Finance, Hearingp Before the Committee on Finance 91st Congress Second session on HTR. lH7OST* Washington, D.C.: 0.3} Government Printing Office, 1970. U.S. Department of Labor, Benefit Series Service Unemploy- ment Insurance. Washington, D.C.: U.S. Government Printing Office. U.S. Department of Labor, Employment and Wages. Washington D.C.: U.S. Government r n ng ice. U.S. Department of Labor, Handbook of Unemployment In- surance Financial Data. WaShington, D.C.: U.Sl Government Printing Office. U.S. Department of Labor, Unemployment Insurance Statistics. Washington,’D.C}: ~U.S. Government Printing’Office. "IIllllflillilflllfl'ifllfiil“