"Arts, Faculty of"@en . "Vancouver School of Economics"@en . "DSpace"@en . "UBCV"@en . "Hanvelt, Robin Alden"@en . "2010-03-22T22:26:22Z"@en . "1980"@en . "Doctor of Philosophy - PhD"@en . "University of British Columbia"@en . "In this study, I examine the influence of unemployment insurance benefits on labour force participation, employment, and unemployment. Conclusions are developed concerning the consequences of the 1971 revision of the Canadian unemployment insurance programme, which differ from those of earlier writings in this field.\r\nMy model estimates the proportions of the population in each labour force state (employment, unemployment, and \"not in the labour force\"). Each labour force state proportion is modelled as a function of the gross flows between the labour force states. This model resembles a Markov model and is similar to the model developed by Toikka (197 6). The decisions by employers and employees that generate the gross flows between labour force states are modelled as behavioural functions of economic variables.\r\nUnlike other studies, this study imposes strict consistency between equations due to the conservation of the population in the gross flows. Other studies have tended to be single equation models and the specification of the equations between studies and in one case, within a study, is not consistent.\r\nThe model is estimated for ten age-sex populations. It is estimated using monthly' data for the period 1961 to 1975. The estimation method is Full Information Maximum Likelihood. Because the system of three equations is singular, one equation is redundant and may be dropped during estimation. Estimation is independent of which equation is dropped.\r\n\r\nThis study brings evidence to support the position that different groups respond in different ways to changes in unemployment insurance. According to the model, prime age men are unresponsive to short-term fluctuations in incentives. Young and old men appear to reduce their labour supply when unemployment insurance benefits are increased. This is the net effect of changes in the gross flows between labour force states. The model suggests that the net labour force participation of women increases in response to increases in unemployment Insurance benefits. Men and women differ in their response to unemployment insurance in two additional ways. First estimated responses for women are generally greater than those for men. While women respond seasonally and non-seasonally to unemployment insurance, the response by men tends to be restricted to seasonal behaviour.\r\nThese findings are consistent with earlier findings in that they suggest a general increase in unemployment and labour force participation due to increases In unemployment insurance. Although my findings suggest some unemployment insurance\u00E2\u0080\u0094induced quit behaviour, they do not suggest a decline in the aggregate level of employment. The dominant result in this study is that unemployment insurance Induces labour force participation, which places upward pressure on employment and unemployment."@en . "https://circle.library.ubc.ca/rest/handle/2429/22263?expand=metadata"@en . "UNEMPLOYMENT INSURANCE AND THE DISTRIBUTION OF WORKERS BETWEEN LABOUR FORCE STATES by Robin Alden Hanvelt B.A., The U n i v e r s i t y of C a l i f o r n i a , 1968 M.A., The U n i v e r s i t y of B r i t i s h Columbia, 1971 A THESIS SUBMITTED IN PARTIAL FULFILMENT OF THE REQUIREMENTS FOR THE DEGREE OF Doctor of Philosophy i n THE FACULTY OF GRADUATE STUDIES (Department of Economics) We accept t h i s t h e s i s as conforming to the required standard THE UNIVERSITY OF BRITISH COLUMBIA June 1980 \u00C2\u00A9 Robin Alden Hanvelt, 1980 In p resent ing t h i s t h e s i s i n p a r t i a l f u l f i l m e n t of the requirements f o r an advanced degree at the U n i v e r s i t y of B r i t i s h Columbia, I agree t ha t the L i b r a r y s h a l l make i t f r e e l y a v a i l a b l e f o r re fe rence and study. I f u r t h e r agree that permiss ion f o r ex tens i ve copying of t h i s t he s i s f o r s c h o l a r l y purposes may be granted by the Head of my Department or by h i s r e p r e s e n t a t i v e s . I t i s understood t ha t copying or p u b l i c a t i o n of t h i s t he s i s f o r f i n a n c i a l gain s h a l l not be a l lowed without my w r i t t e n permi s s ion . Department nf Economics The U n i v e r s i t y of B r i t i s h Columbia 2075 Wesbrook P lace Vancouver, Canada V6T 1W5 Date 21 August 1980 - i -ABSTRACT In t h i s study, I examine the influence of unemployment insurance be n e f i t s on labour force p a r t i c i p a t i o n , employment, and unemployment. Conclusions are developed concerning the consequences of the 1971 r e v i s i o n of the Canadian unemployment insurance programme, which d i f f e r from those of e a r l i e r writings i n t h i s f i e l d . My model estimates the proportions of the population i n each labour force state (employment, unemployment, and \"not i n the labour f o r c e \" ) . Each labour force state proportion i s modelled as a function of the gross flows between the labour force states. This model resembles a Markov model and i s s i m i l a r to the model developed by Toikka (197 6). The decisions by employers and employees that generate the gross flows between labour force states are modelled as behavioural functions of economic v a r i a b l e s . Unlike other studies, t h i s study imposes s t r i c t consistency between equations due to the conservation of the population i n the gross flows. Other studies have tended to be s i n g l e equation models and the s p e c i f i c a t i o n of the equations between studies and i n one case, within a study, i s not consistent. The model i s estimated for ten age-sex populations. It i s estimated using monthly' data f o r the period 1961 to 1975. The estimation method i s F u l l Information Maximum L i k e l i h o o d . Because the system of three equations i s s i n g u l a r , one equation i s redundant and may be dropped during estimation. Estimation i s independent of which equation i s dropped. - i i -O This study brings evidence to support the p o s i t i o n t h a t : d i f f e r e n t groups respond i n d i f f e r e n t ways to changes i n unemployment insurance. According to the model, prime age men are unresponsive to short-term f l u c t u a t i o n s i n incentives. Young and old men appear to reduce t h e i r labour supply when unemployment insurance benefits are increased. This i s the net e f f e c t of changes i n the gross flows between labour force s t a t e s . The model suggests that the net labour f o r c e p a r t i c i p a t i o n of women increases i n response to increases i n unemployment Insurance b e n e f i t s . Men and women d i f f e r i n t h e i r response to unemployment insurance i n two a d d i t i o n a l ways. F i r s t estimated responses f o r women are generally greater than those f o r men. While women respond seasonally and non-seasonally to unemployment insurance, the response by men tends to be r e s t r i c t e d to seasonal behaviour. These find i n g s are consistent with e a r l i e r f i n d i n g s i n that they suggest a general increase i n unemployment and labour force p a r t i c i p a t i o n due to increases In unemployment insurance. Although my fin d i n g s suggest some unemployment insurance\u00E2\u0080\u0094induced qui t behaviour, they do not suggest a d e c l i n e i n the aggregate l e v e l of employment. The dominant r e s u l t i n t h i s study i s that unemployment insurance Induces labour for c e p a r t i c i p a t i o n , which places upward pressure on employment and unemployment. - i i i -TABLE OF CONTENTS Page ABSTRACT i TABLE OF CONTENTS i i i LIST OF TABLES v i LIST OF FIGURES v i i ACKNOWLEDGEMENTS v i i i CHAPTER 1 INTRODUCTION Objective of the Study 1 Changes i n the Unemployment Insurance Programme 2 Ben e f i t s . 3 Benefit Period 4 Labour Market Theories 5 Economic and Demographic Changes..... \u00E2\u0080\u00A2 7 Summary of Findings. 8 Footnotes.... 8 2 LITERATURE REVIEW 9 A S e l e c t i v e L i t e r a t u r e Review 9 Footnotes 24 3 THEORETICAL APPROACH 25 S o c i a l Roles. 26 Demographic Patterns 28 S e n i o r i t y on the Job 29 - i v -CHAPTER Page 3 THEORETICAL APPROACH Con't Segmented Labour Markets. 29 Accounting I d e n t i t i e s \u00E2\u0080\u00A2 \u00E2\u0080\u00A2\u00E2\u0080\u00A2\u00E2\u0080\u00A2\u00E2\u0080\u00A2 30 Modelling - The General Approach .. 32 The determinants of ( i j ) t + i 33 The determinants of the Six Component P r o b a b i l i t i e s 36 Layoffs and Job Offers 37 Quits and Job Acceptance 39 Labour Force Entry and Withdrawal 41 Component P r o b a b i l i t i e s Combined into the Model..... 43 Footnotes \u00E2\u0080\u00A2 47 4 ESTIMATION METHOD AND DATA 48 S p e c i f i c a t i o n of the Model f o r Estimation 48 Formulation of the L i k e l i h o o d Function....... 48 Data. 50 5 RESULTS 56 E m p i r i c a l Analysis - Assuming Constant T r a n s i t i o n P r o b a b i l i t i e s 56 Dynamic Properties of the \"Null Hypothesis Model\" 59 Em p i r i c a l Analysis - Economic Model 62 -v-CHAPTER Page 5 RESULTS Con't Testing the S p e c i f i c a t i o n 7 7 Analysis of the Economic Model.... 82 Decisions by Workers Employed 90 Decisions by Workers Unemployed 94 Decisions by Persons Not i n the Labour Force 99 Estimates of Average E l a s t i c i t i e s of Responses to Changes i n Unemployment Insurance 103 Conclusions 105 Footnotes I l l BIBLIOGRAPHY 112 \ - v l -LIST OF TABLES TABLE Page 1 Estimated Constant Transition Probabilities... 57 Estimated Economic Model without Seasonal Dummy 2.1 Men, Dependent Variable: E t + i / P t + j . 63 2.2 Men, Dependent Variable: U t+i/P t +i ..... 64 2.3 Men, Dependent Variable: N t + 1/P t + 1. 65 2.4 Women, Dependent Variable: Et+l/ pt+l\u00E2\u0080\u00A2 \u00E2\u0080\u00A2 * 66 2.5 Women, Dependent Variable: U t+i/P t.fi 67 2.6 Women, Dependent Variable: N t4.j/P t +j 68 Estimated Economic Model with Seasonal Dummy Variables 3.1 Men, Dependent Variable: Et+l/ pt+l 7 1 3.2 Men,-Dependent Variable: Ut+l/ pt+l \u00E2\u0080\u00A2\u00E2\u0080\u00A2 7 2 3.3 Men, Dependent Variable: N t + i / P t + i 73 3.4 Women, Dependent Variable: E t + l / p t + l * * * * 7 <^ 3.5 Women, Dependent Variable: Ut+i/Pt+1 7 ^ 3.6 Women, Dependent Variable: Nt4.i/Pt+1 76 4.1 Ln of the Likelihood Functions - Men 78 4.2 Ln of the Likelihood Functions - Women...... 79 5.1 Likelihood Ratio Tests - Men 80 5.2 Likelihood Ratio Tests - Women 81 6 Summary of Economic Model Estimates 83 - v i i -LIST OF FIGURES FIGURE Page 1 Labour Force D i s t r i b u t i o n of the Population. 21 2 Labour Force Flow Schematic... 25 - v i i i -ACKNOWLEDGEMENTS In the preparation of any study, the encouragement and i n t e l l e c t u a l support of others i s e s s e n t i a l . S p e c i a l acknowledgement and thanks go to my d i s s e r t a t i o n committee c o n s i s t i n g of Professors G.C. Archibald, P. Chinloy, and A. Woodland. I have also benefited from the suggestions and comments of Professors C. Boonekamp, J . Cragg, J . Kesselman, C. R i d d e l l , L. Robb, and R. Shearer. I am a l s o indebted to Professor G. Rosenbluth f o r h i s consistent support and encouragement during periods when they were of the greatest b e n e f i t to me. The completion of t h i s t hesis was a s s i s t e d by the e x c e l l e n t typing of Miss C. Rogacz who typed successive d r a f t s i n time to meet deadlines. F i n a l l y , I would never have completed the t h e s i s without the continuous support and s a c r i f i c e of my family. This thesis i s dedicated to my wife, Vera, and my sons, Marc and Jonathan. CHAPTER 1 - INTRODUCTION OBJECTIVE OF THE STUDY In t h i s study, I examine the influence of unemployment insurance b e n e f i t s on labour f o r c e p a r t i c i p a t i o n , employment, and unemployment. The study's focus i s on the supply of labour, which has been the subject of recent controversy. Conclusions are developed concerning the consequences of the 1971 r e v i s i o n of the Canadian unemployment, insurance programme, which d i f f e r from those of e a r l i e r writings i n t h i s f i e l d . Unemployment Insurance i s an income t r a n s f e r programme. Therefore i t i s subject to the same concerns that surround other Income t r a n s f e r schemes. The issue which generally a t t r a c t s the greatest p u b l i c i n t e r e s t i s whether income t r a n s f e r schemes encourage i d l e n e s s , thus reducing employment and the gross n a t i o n a l product. This controversy can be traced back to the beginning of income r e l i e f . In the case of Income t r a n s f e r schemes which r e q u i r e work h i s t o r i e s as part of t h e i r determination of e l i g i b i l i t y ( l i k e unemployment Insurance), the question of induced labour f o r c e p a r t i c i p a t i o n i s , however, a l s o Important. This phenomenon may o f f s e t the induced Idleness expected by some. A d d i t i o n a l l y , an income support programme such as unemployment insurance may draw o f f r e c i p i e n t s of other kinds of income support depending upon t h e i r r e l a t i v e a t t r a c t i v e n e s s . It can a l s o be viewed as providing subsidies to employers thus a f f e c t i n g t h e i r labour market behaviour. The extent to which employers and employees adjust t h e i r labour -2-market behaviour in order to benefit from such schemes w i l l depend upon the size of the potential individual gains. It w i l l also depend upon the level of discretion of those persons able to qualify for the income transfer. To the extent that workers and/or employers can adjust their circumstances, we might expect to find significant -responses. Finally, larger changes in an income transfer scheme would tend to affect the decisions of workers and employers more than smaller changes. This study w i l l not attempt to deal with a l l these issues. It i s restricted to examining the influence of unemployment Insurance on the distribution of the population between labour force states. The time period examined is 1961 to 1975 which includes the major change in the unemployment insurance programme i n 197 1. Before we can discuss the issues to be investigated, the details of the 1971 programme changes are required. Changes in the Unemployment Insurance Programme. In 1971, the Parliament of Canada changed the Unemployment Insurance (U.I.) Act significantly. The changes were apparently the result of a s h i f t in philosophy from a programme restricted to 'low-income' earners to a generalized programme with nearly universal coverage. It was estimated that under the new scheme 1.6 million additional employees* would have been covered by U.I. in 1968. That i s approximately 16% of the labour force in 1968. The new scheme would have covered approximately 96% of employees in the labour force i n 1968. - 3 -Prior to the 1971 changes, a person had to have earnings less than $7,800 per year to be Insured. He had to meet four conditions in order to qualify for benefits: 1) he had to be unemployed defined as an Interruption of earnings, 2) he had to be capable of working, 3) he had to be unable to find appropriate employment, and 4) he had to have contributed to the programme for at least 30 weeks during the last two years with at least 8 weeks of contributions during the last year. Casual employees were not covered by U.I. Under the new programme the annual income limitation of $7,800 was l i f t e d . Persons with interrupted earnings for reasons of maternity, retirement, or sickness were el i g i b l e for special benefits. A person with as l i t t l e as 8 weeks of contributions during the last year became el i g i b l e for benefits. A contribution week required earnings of at least $25 during the week. This provision extended coverage to qualifying casual employees. Benefits Before 1971 unemployment insurance, benefits were based on the person's earnings level, length of time in the labour market, and the number of dependents. These benefits were an average 43 per cent^ of earnings with a maximum benefit level of $53 per week. Under the new scheme, benefits are set at two-thirds of earnings with a new maximum benefit of $100 per week. Individuals with extended periods of unemployment became e l i g i b l e for benefits of 75 per cent of their earnings up to $100 i f they had dependents. The new scheme provided for maternity benefits during the period that women -4-were previously disqualified from benefits. Persons who lost their job due to illness became e l i g i b i l e for special benefits. Previously persons unable to work due to illn e s s were only e l i g i b l e for benefits i f they were already receiving benefits before becoming i l l . Finally, persons between 65 and 70 years old opting for the Canada or Quebec Pension Plan benefits were e l i g i b l e for a three-week retirement benefit. Benefit Period Prior to 1971 two types of benefits were available. Seasonal benefits were available for up to 23 weeks between December 1 and May 15 based upon employment prior to the previous March 31. Thirteen weeks of seasonal benefits were available after fifteen weeks of work. Five-sixths of a week of benefit e l i g i b i l i t y was earned for each additional week up to an additional ten weeks of benefits. Regular benefits were somewhat less generous. A minimum of th i r t y weeks of work were required for fifteen weeks of benefit e l i g i b i l i t y . An additional week of benefit e l i g i b i l i t y was earned for each additional two weeks of work. . 0 After the 1971 changes to the U.I. programme, a person could pass through five phases of benefits not to exceed fifty-one weeks of benefits in total. For persons with at least twenty weeks of work, three weeks of Phase 1 benefits were available after a two week waiting period. For persons with at least eight weeks of work and/or those completing Phase 1 benefits, Phase 2 benefits were available* Eight to fifteen weeks of work earned eight benefit weeks of -5-e l i g i b i l i t y . Each a d d i t i o n a l week of employment over f i f t e e n added a week of i e l i g i b i l i t y up to nineteen benefit weeks. For persons completing Phase 2, up to eighteen a d d i t i o n a l benefit weeks of e l i g i b i l i t y were added i n Phase 3 depending upon the national unemployment rate. Persons who had completed Phase 3 and had twenty or more weeks of work were e l i g i b l e f o r Phase 4. In Phase 4 the person was e l i g i b l e f o r one benefit week for every two employment weeks i n excess of twenty employment weeks plus an a d d i t i o n a l two b e n e f i t weeks. The maximum e l i g i b i l i t y under Phase 4 was eighteen be n e f i t weeks. F i n a l l y , those who complete Phase 4, could e s t a b l i s h e l i g i b i l i t y f o r up to an a d d i t i o n a l eighteen benefit weeks i f t h e i r r e g i o n a l unemployment rate was i n excess of four per cent and exceeded the n a t i o n a l unemployment rate. Total e l i g i b i l i t y from a l l Phases of the programme was r e s t r i c t e d to f i f t y - o n e weeks. Labour Market Theories A knowledge of the views of the labour market and unemployment held by advocates and opponents of income t r a n s f e r schemes i s e s s e n t i a l i n order to understand the controversy following the 1971 changes to the U.I. programme. The advocates, i n c l u d i n g the Honourable Bryce Mackasey, M i n i s t e r of Manpower and Immigration who introduced the changes, concentrated on unemployment as a consequence of the employer's decisions. They adopted the Keynesian deficient-demand perspective which views unemployment as a consequence of l a y o f f s and f i r i n g s i n response to d e c l i n i n g economic a c t i v i t y . This unemployment i s viewed as involuntary because the unemployed can not d i r e c t l y -6-a f f e c t the f a c t o r s determining the demand f o r labour. In the white paper on Unemployment Insurance i n the 70's (1970), the government a l s o focused on unexpected l a y o f f s r e s u l t i n g from automation and other t e c h n o l o g i c a l changes. The c r i t i c s of the 1971 programme changes concentrated on the p o t e n t i a l labour supply responses to the new programme. The simplest view i s based on a work/leisure tradeoff model In which the workers are w i l l i n g to supply varying amounts of labour at various wage l e v e l s . From t h i s point of view unemployment insurance b e n e f i t s subsidize l e i s u r e thus p o t e n t i a l l y causing employees to reduce t h e i r work e f f o r t . I t i s c l e a r that i f a person Is i n a s i t u a t i o n where he can a c t u a l l y trade o f f work against l e i s u r e he may make these adjustments. A s l i g h t l y more so p h i s t i c a t e d v e r s i o n of t h i s model introduces a t h i r d a c t i v i t y , search f o r a job or a better job, that can be traded o f f against e i t h e r work or l e i s u r e . To ask which of these models i s the more appropriate i s equivalent to asking how much d i s c r e t i o n employees have, how many are already under-employed, and whether we are l i k e l y to f i n d them adjusting t h e i r short-term behaviour i n response to changes i n an income subsidy programme l i k e unemployment Insurance. In Canada, the choice of an appropriate model i s f u r t h e r complicated by the f a c t that the economy has a major seasonal sector i n a d d i t i o n to the non-seasonal sector. For the f i s h i n g , f o r e s t r y , a g r i c u l t u r e , t r a n s p o r t a t i o n , and tourism sectors we know that the demand f o r labour has a strong seasonal pattern. Some employees i n these sectors can earn s u f f i c i e n t income -7-f o r the e n t i r e year In a few months of seasonal work. Others such as the tree planters, the f i s h plant workers, the farm workers, the truck 'swampers', the waitresses, or other low paid employees must have other sources of Income. Employment income i s scarce by d e f i n i t i o n In the 'off season*. Any u s e f u l i n v e s t i g a t i o n of the e f f e c t s of unemployment insurance must take seasonal f a c t o r s into account. It i s a l s o necessary to d i s t i n g u i s h the responses of 'primary* and 'secondary' wage earners In a family. Although a growing number of i n d i v i d u a l s do not conform to t r a d i t i o n a l family r o l e s , these r o l e s are s t i l l very important i n the t o t a l p i c t u r e . Primary earners t y p i c a l l y can not a f f o r d l e i s u r e or job search even when wages are low or unemployment benefits high. They w i l l not give up job advancement or career development f o r the short-term b e n e f i t of higher unemployment insurance payments. In seasonal i n d u s t r i e s , however, they may 'conspire' with employees to accept longer periods of seasonal l a y o f f when unemployment benefits are higher. Secondary workers, on the other hand, having t r a d i t i o n a l rewarding r o l e s to play outside the labour force, may leave i t i n response to quite small and temporary changes i n Incentives. Economic and Demographic Changes Since the 1971 changes to unemployment Insurance, the l e v e l and r a t e of unemployment have increased dramatically i n Canada. This has a l s o been associated with higher rates of i n f l a t i o n compared with the previous decades. The generation of the post-war * baby-boom' began reaching the age of labour f o r c e entry In the l a t e 1960's with the -8-edge of the boom at the age of 25 years o l d l n 1971. This study c o n t r o l s f o r these associated phenomena which, we would expect, have a l s o had a s i g n i f i c a n t e f f e c t on the labour market. Summary of Findings This study brings evidence to support the p o s i t i o n that d i f f e r e n t groups responded to the 1971 U.I. changes i n d i f f e r e n t ways. The r e s u l t s on prime age men suggest that t h i s group i s not responsive t o short-term f l u c t u a t i o n i n inc e n t i v e s l i k e U.I. Younger and older men are responsive to the U.I. changes. Women ge n e r a l l y are s e n s i t i v e to short-terra f l u c t u a t i o n i n incentives. The estimated net e f f e c t of a small increase i n unemployment insurance b e n e f i t s induces women to increase t h e i r labour force p a r t i c i p a t i o n . This Increases the proportions of women employed and unemployed. The r e s u l t a n t e f f e c t i s an increase i n the unemployment rate f o r women. The estimated net e f f e c t on the labour market behaviour of men i s a small general d e c l i n e i n labour force p a r t i c i p a t i o n and the proportion o f men unemployed when unemployment be n e f i t s increase. The proportion of prime age men employed increases while only the youngest and oldest men appear to reduce employment. The r e s u l t a n t e f f e c t i s a general decline i n the unemployment rate f o r men. Footnotes 1) See Unemployment Insurance i n the 70's pp. 10-12. 2) See Unemployment Insurance i n the 70's p. 10. - 9 -CHAPTER 2 LITERATURE REVIEW There i s no one way to examine the labour market Issue r a i s e d a f t e r the 1971 changes to the Unemployment Insurance Act. I review studies which use three d i f f e r e n t types of data each of which contains Important l i m i t a t i o n s . The studies using aggregate data attempt to provide estimates of the t o t a l Impact of the U.I. changes. There i s some doubt that they have separated the demand and supply e f f e c t s o f unemployment insurance. Other studies use disaggregated data ( i . e . , they p a r t i t i o n the population i n t o groups by age and sex). These studies are b e t t e r able to i s o l a t e the supply e f f e c t s of unemployment insurance. The t h i r d type of study uses l o n g i t u d i n a l data ( i . e . , i n d i v i d u a l time-series data). The one study that uses t h i s type of data assumes a l l unemployment i s voluntary ( a l l supply e f f e c t ) and the data sample excludes a p o t e n t i a l l y important group. My own study uses disaggregated data p a r t i t i o n i n g the population i n t o age-sex groups. The r e s u l t s are presented f o r each group. The a study i s focused on the expected di f f e r e n c e s of supply behaviour between these groups. A S e l e c t i v e L i t e r a t u r e Review . Grubel, Maki, and Sax (1975a) published one of the e a r l i e s t studies of unemployment insurance and the labour supply i n Canada. They hypothesize that increased unemployment insurance b e n e f i t s induce workers to q u i t t h e i r jobs because the cost of l e i s u r e and job search -10-(while unemployed) i s reduced. Their approach i s based upon a work/leisure tradeoff model. They also hypothesize that workers \" i n v o l u n t a r i l y \" unemployed w i l l extend t h e i r period of unemployment by reducing t h e i r job search e f f o r t s when benefits r i s e . F i n a l l y , they acknowledge that persons \"not i n the labour f o r c e \" may be encouraged to enter the labour force and search f o r work of \"short duration\" i n order to q u a l i f y f o r unemployment insurance b e n e f i t s . I n t e r e s t i n g l y , the authors do not attempt to estimate the increase i n p a r t i c i p a t i o n rates that they a n t i c i p a t e i n t h e i r t h e o r e t i c a l discussion. Nonetheless they estimate an annual time-series model f or the period 1953 to 1972. The log of the aggregate unemployment rate i s modelled as dependent upon the r a t i o of the average unemployment insurance b e n e f i t to the average weekly wage (\"replacement wage\"), the percentage change i n c u r r e n t - d o l l a r gross n a t i o n a l product ( f or the current year and lagged one year), the female and male p a r t i c i p a t i o n rates, and the percentage of new and renewal claims for unemployment benefits which are ru l e d i n e l i g i b i l e during the period (adjusted f o r c y c l i c a l v a r i a t i o n ) . The authors conclude that the l i b e r a l i z a t i o n of unemployment insurance may have increased the measured unemployment rate by 0.8 percentage points i n 197 2. This conclusion i s on a very weak s t a t i s t i c a l foundation since i t i s based upon one annual observation a f t e r the changes. The replacement r a t i o v a r i a b l e i s v i r t u a l l y constant from 1953 u n t i l 1971 varying between 0.24 and 0.31, and then increases to 0.41 In 1972.1 The variable was also constructed without adjusting for the changes in the Canadian Income Tax Act In 1972 when unemployment insurance benefits became taxable for the f i r s t time (Kaliski, 1975). When Grubel et a l . examined the taxation implications for their model, they found that assuming average tax rates of 10% and 20% reduced their estimated effect of the 1972 changes on the measured unemployment rate to 0.5 and 0.0 percentage points respectively (Grubel, Makl, and Sax, 1975b).^ The authors declared their results \"exploratory\" which I consider accurate for several reasons. F i r s t l y , they never examined the impact on participation rates of changes in the unemployment insurance programme. Secondly, consideration of a l l the possible effects of the changes to the Unemployment Insurance programme on labour supply decisions would require more explicit modelling of the operation of the labour market. Finally, the virtual collapse of their main result when unemployment insurance benefits for 1972 are adjusted for taxation undermines their conclusions and policy recommendations. Green and Cousineau (1976) prepared a study for the Economic Council of Canada on the impact of unemployment insurance on unemployment in Canada. They f i r s t examine the unemployment rate/job vacancy pattern for 1951 to 1973. They apply the models of Gujurati (1972) and Foster (1973) which model the log of the unemployment rate as a function of the log of the rate of vacancies, and a time\u00E2\u0080\u00A2trend (Gujurati) or the log of the lagged unemployment rate (Foster). In - 1 2 -each case they f i n d that the estimated unemployment rate i s l e s s than the observed l e v e l s f o r 1972-73. They i n t e r p r e t t h i s f i n d i n g as evidence that a s t r u c t u r a l s h i f t occurred i n 1972 and 1973. They demonstrate t h i s s h i f t v i s u a l l y using a graph of the unemployment rate and the index of the vacancy rate. The \" s h i f t \" they observe i s not very d i f f e r e n t from a \" s h i f t \" i n 1954-56, and there i s another i n t e r e s t i n g \" s h i f t \" i n 1969. The 1972-73 s h i f t would be consistent with a recovery from the 1971-72 recession concurrent with a s i g n i f i c a n t increase i n the p a r t i c i p a t i o n rate. Unfortunately these \" s h i f t s \" are l a r g e l y i n the eye of the beholder, and without e x p l i c i t modelling, t h e i r cause remains untested. Green and Cousineau also construct a time-series model of the determinants of the number of unemployed workers and the labour force p a r t i c i p a t i o n rate. The number of unemployed workers i s expressed as a f u n c t i o n of the deviations of the gross national product from i t s trend (lagged three quarters),, the deviations of the labour force from i t s trend, the long-term labour force trend, the r a t i o of the average unemployment insurance benefits to the average weekly wage, and the rate of r e f u s a l of unemployment insurance claims weighted by the proportion of the labour force covered by the unemployment insurance programme. The p a r t i c i p a t i o n r a t e i s modelled as a function of the deviations of the average weekly wage from i t s long-term trend (lagged two q u a r t e r s ) , the deviations of the gross n a t i o n a l product from i t s long-term trend (lagged two quarters), the b i r t h rate, and a -13-tlme trend. Green and Couslneau assume that the d i f f e r e n c e between observed and predicted labour force p a r t i c i p a t i o n resulted from the changes i n the Unemployment Insurance Act. This lack of e x p l i c i t modelling of the impact of unemployment insurance on the p a r t i c i p a t i o n r a t e generates very i n t e r e s t i n g problems i n i n t e r p r e t i n g t h e i r r e s u l t s . They estimate the \" d i r e c t e f f e c t \" of the 1971 changes i n unemployment insurance on the number of unemployed workers to be an increase of 56,788 i n 1972 and 32,880 i n 1973. These estimates are based upon changes i n the two unemployment insurance v a r i a b l e s i n the unemployment equation. They then estimate an increase i n the labour f o r c e of 13,000 i n 1972, and 129,000 i n 1973 r e s u l t i n g from the unemployment insurance changes. These estimates are based upon the d e v i a t i o n between the predicted and act u a l values i n the p a r t i c i p a t i o n rate equation. Using the estimated c o e f f i c i e n t f o r the labour force i n the unemployment equation, they estimate the \" i n d i r e c t e f f e c t \" on the number of unemployed to be 3,250 i n 197 2 and 32,250 i n 197 3. They thus c a l c u l a t e the t o t a l e f f e c t as an increase i n the number of unemployed workers by 60,038 i n 1972 and 65,130 i n 1973. 3 Although the authors express i n t e r e s t i n unemployment e x c l u s i v e l y , t h e i r r e s u l t s f o r unemployment and labour f o r c e p a r t i c i p a t i o n imply changes i n employment. Their \" d i r e c t e f f e c t \" on unemployment i s t h e i r estimate of the decline i n employment due to unemployment insurance. However the increase i n labour force p a r t i c i p a t i o n due to unemployment -14-Insurance which did not add to unemployment must have added to employment. When we combine the \"direct effect\" and \"indirect effect\" on employment, we find that Green and Cousineau's model estimates a decline of 47,038 in 1972 and an increase of 63,870 In 1973. This interesting f l i p - f l o p would suggest that the model may not be very stable or that the delayed unemployment insurance effect on labour force participation swamps the immediate \"direct effect\" on employment. If the latter i s accepted, the result directly contradicts the conclusion by Grubel et a l . (1975a) that unemployment Insurance reduced \"society's aggregate output of market goods\".V Two further comments are warranted. Like Grubel, Maki, and Sax, Green and Cousineau did not adjust the unemployment insurance benefits after 1971 for the changes In the income tax legislation. Secondly, like Grubel et a l . , Green and Cousineau use a relatively simple model to capture relatively complex decisions made by labour force members. Samuel Rea (1977) studied the potential effects of the Canadian Unemployment Insurance programme changes on labour supply using the Unemployment Insurance Commission's Historical Data Base.5 This study i s restricted to individuals for whom an unemployment insurance contribution record and an income tax f i l e existed between the years 1966 and 1970. Consequently, individuals Induced into the labour force by the changes in the Unemployment Insurance programme may not be included. Equally, while the 1971 programme changes expanded coverage to v i r t u a l l y a l l workers, workers previously excluded from -15-unemployment insurance would a l s o be excluded from the data base. Rea estimates the labour supply using a work/leisure model. He assumes that workers have a one year d e c i s i o n horizon, and that a l l unemployment experienced by the sample i s voluntary. He models weeks worked per year as a fu n c t i o n of the r e a l non-wage income, the r e a l weekly earnings, the percentage change i n p r o v i n c i a l employment, the p r o v i n c i a l unemployment r a t e , and other c o n t r o l dummy v a r i a b l e s f o r Province, m a r i t a l status, dependents, age, year, and sex. Rea concludes that the workers i n h i s sample would reduce t h e i r weeks worked i n response to the 1971 changes to Unemployment Insurance i n Canada. However he c o r r e c t l y points out that i f he relaxes h i s assumption that a l l unemployment i s voluntary, persons i n v o l u n t a r i l y unemployed may replace those that v o l u n t a r i l y reduce t h e i r work e f f o r t . This could r e s u l t i n no change to measured unemployment. Swan, Steinberg, and MacRae (1976) studied the e f f e c t s of Income maintenance programmes ( i n c l u d i n g unemployment insurance) on labour supply i n the three Maritime provinces and i n Canada as a whole. They regress the p a r t i c i p a t i o n rate f o r each region on the maximum unemployment insurance d o l l a r b e n e f i t s d i v i d e d by the minimum number of q u a l i f y i n g weeks, the number of weeks of work required to q u a l i f y f o r an a d d i t i o n a l benefit week, and a number of c o n t r o l v a r i a b l e s ( i n c l u d i n g the b i r t h rate, the Canadian wage r a t e , the Canadian unemployment rate, a time trend, and q u a r t e r l y seasonal dummy v a r i a b l e s ) . Their r e s u l t s lead them to three conclusions. \" F i r s t , the greater generosity of the unemployment insurance system has not decreased true labour force p a r t i c i p a t i o n i n the Maritimes. Second, p a r t i c i p a t i o n has increased i n New Brunwick and Prince Edward Island, and more so than f o r Canada as a whole. T h i r d , p a r t i c i p a t i o n has not decreased i n Nova Scotia\".** There are two points worthy of note regarding these r e s u l t s . F i r s t l y , the authors do not mention any adjustment to the unemployment insurance b e n e f i t s to take into account the changes i n the tax status of b e n e f i t s . Secondly, the model i m p l i c i t l y assumes that the two unemployment insurance v a r i a b l e s are the appropriate d e c i s i o n parameters f o r workers deciding to stay i n the labour f o r c e or t o enter the labour force, to quit employment, and to accept employment. I would a n t i c i p a t e that workers q u i t t i n g or accepting employment are weighing the returns to unemployment, while persons outside the labour f o r c e are weighing the a l t e r n a t i v e returns to market and non-market a c t i v i t y . The v a r i a b l e s used by the authors are not included e x p l i c i t l y to capture these d e c i s i o n processes, but are considered measures of an abstract concept c a l l e d the \"degree of generosity\". A more e x p l i c i t modelling of the operation of the labour market and worker decisions would be preferred. P a r t i c i p a t i o n rates were also examined f o r t h e i r s e n s i t i v i t y t o unemployment insurance by S h a r i r and Kuch (1977). The authors estimated equations f o r various age-sex groups. Their model makes the p a r t i c i p a t i o n rate a f u n c t i o n of the employment/population r a t i o f o r -17-the p a r t i c u l a r age-sex group under study, the employment/population r a t i o f o r a l l other age-sex groups, a time trend, and two unemployment Insurance v a r i a b l e s f o r the 1955 and 1971 programme changes. The unemployment insurance v a r i a b l e s were constructed to capture the e f f e c t s of Increased coverage and b e n e f i t s , and the increase i n the waiting period f o r l a y o f f s . Generally, the authors concluded that the increase i n unemployment insurance benefits increased labour f o r c e p a r t i c i p a t i o n f o r most age sex groups. These authors a l s o examined the p o t e n t i a l impact of the unemployment insurance programme on temporary l a y o f f s . This work i s p r i m a r i l y a response to work by M. F e l d s t e i n (1976,1978) who has shown that most persons l a i d o f f i n the U.S. return to work with t h e i r previous employer. Feldstein's work suggests that there are good grounds - to b e l i e v e that unemployment insurance b e n e f i t s i n the U.S. have increased temporary l a y o f f s . The argument i s based on an \" i m p l i c i t contract\" between workers and employers i n seasonal Industries who use the unemployment insurance b e n e f i t s to subsidize the seasonal a c t i v i t y . This i s l e s s l i k e l y to occur i f \"experience r a t i n g \" i s introduced to adjust the employers' premiums. Experience r a t i n g has not been introduced i n Canada. Sharir and Kuch model a simple approach to l a y o f f s . They regard seasonally adjusted temporary l a y o f f s (30 day or l e s s ) per head of population as a fu n c t i o n of seasonally adjusted t o t a l employment per head of population, a time trend, the r a t i o of unemployment insurance -18-b e n e f i t s to average weekly wage, the unemployment Insurance waiting period f o r l a i d - o f f workers, and the proportion of the labour force covered by unemployment insurance. Their model suggests that temporary l a y o f f s increased a f t e r the 1955 programme changes and that the 1971 programme changes had a s t a t i s t i c a l l y i n s i g n i f i c a n t e f f e c t on temporary l a y o f f s . This i n v e s t i g a t i o n i n t o l a y o f f s represents a more f r u i t f u l s trategy than the e a r l i e r papers discussed, where very complex movements between labour force states were not e x p l i c i t l y modelled. Although S h a r i r and Kuch's work i s l i m i t e d to one type of movement between the two labour force s t a t e s , employment and unemployment, i t focuses a t t e n t i o n on a labour market process which i s disguised by the net changes i n labour force s t a t e s . I t i s u n l i k e l y that unemployment Insurance a f f e c t s workers' decisions the same way when they are i n d i f f e r e n t labour f o r c e s t a t e s . Therefore i t would be u s e f u l to examine the labour market decisions of workers i n d i f f e r e n t labour force s t a t e s , such as l a y o f f s due t o i m p l i c i t c o n t r a c t i n g . There have been attempts among advocates of job search theory to examine the t h e o r e t i c a l i m p l i c a t i o n s of unemployment Insurance f o r job search. The theory (see Ehrenberg and Oaxaca, 1976, Lazar, 1978, Marston, 1975, and Mortensen, 1977) suggests that the t h e o r e t i c a l impact of unemployment insurance on job search v a r i e s depending upon the searcher's s i t u a t i o n . Those unemployed and r e c e i v i n g b e n e f i t s may extend job search because unemployment insurance b e n e f i t s may be -19-seen as a subsidy. Those unemployed and not receiving unemployment Insurance (i.e. exhaustees, or new labour force entrants) may be induced to accept employment sooner i f they w i l l then qualify for unemployment insurance benefits at a later date. Chapin (197 1) produced one of the earliest studies of the impact of unemployment Insurance on the duration of unemployment. He modelled the average actual duration of unemployment insurance claims as a function of the unemployment rate of the insured population, the average unemployment insurance benefit payment divided by the average wage of the last insured employment and the maximum duration of unemployment insurance claims. Empirically he found that the maximum duration of unemployment insurance claims had more Impact on unemployment duration than the benefit wage ratio for equal percentage changes. The most significant variable was the unemployment rate. Marston (1975) working on U.S. data found that there was a significant expected difference between the duration of spells of unemployment for el i g i b l e persons and persons i n e l i g i b l e for unemployment insurance. This Is similar to Chapin's finding that the maximum duration of unemployment insurance claims has a large influence on the duration of unemployment spells. Ehrenberg and Oaxaca (1976) i n their study of U.S. data also found that unemployment insurance extended the spell of unemployment. They had anticipated that an increase i n wages should occur for workers who appeared to extend their job search i n response to unemployment insurance. They -20-did not f i n d any s i g n i f i c a n t change i n wages a f t e r job search. Although the wage i s only one component of the return to employment, the others ( i . e . advancement p o t e n t i a l , working co n d i t i o n s , psychic b e n e f i t s , etc.) are not measured, and some of them can not be measured. Lazar (1977) has produced the only published empirical work on job search using Canadian data. He uses unpublished data supplied by S t a t i s t i c s Canada on a turnover measure (the number of new weekly s p e l l s of unemployment as a proportion of the labour f o r c e ) , and on continuation p r o b a b i l i t i e s (the p r o b a b i l i t y an unemployed i n d i v i d u a l w i l l remain unemployed another week). The continuation p r o b a b i l i t i e s were transformed i n t o a measure of average s p e l l duration i n weeks. He modelled the turnover rates and continuation p r o b a b i l i t i e s as functions of seasonal dummies, the average capacity u t i l i z a t i o n rate f o r t o t a l manufacturing, and an unemployment insurance dummy v a r i a b l e constructed to measure the 1971 changes. Lazar's e m p i r i c a l work suggests that the 1971 changes Increased the duration of a s p e l l of unemployment f o r a l l groups except women 45 years and old e r . He al s o concludes that turnovers f o r women and young men (14-24 years) increased a f t e r 1971. Because Lazar's approach aggregates new s p e l l s of unemployment due to q u i t s , l a y o f f s , and labour f o r c e entry i n t o one turnover measure, the decisions by workers that net out to h i s r e s u l t can not be i d e n t i f i e d . While the above studies have suggested that unemployment, increased and that average duration of s p e l l s of unemployment increased a f t e r the 1971 unemployment insurance changes, none of these studies have determined what sequence of labour market decisions were responsible f o r t h e i r r e s u l t s . Increased unemployment rates and average duration s p e l l s of unemployment could r e s u l t from increased p a r t i c i p a t i o n , \" i m p l i c i t c ontract\" l a y o f f s , q u i t s , extended job search, or a combination of these f a c t o r s . Mortensen's (197 7) model t h e o r e t i c a l l y i d e n t i f i e s these f a c t o r s , but none of the studies above attempt to i d e n t i f y the c o n t r i b u t i o n s of these f a c t o r s . A u s e f u l way of looking at the problem can be i l l u s t r a t e d i n the following f i g u r e . Labour Force D i s t r i b u t i o n of the Population Employment Labour Force Labour Force Population Figure 1 -22-The v e r t i c a l axis represents employment as a proportion of the labour f o r c e and the h o r i z o n t a l a x i s represents the proportion of the population i n the labour force (labour force p a r t i c i p a t i o n r a t e ) . The curve \"cc\" i s a rectangular hyperbola representing the constant proportion of the population employed ( E / p = E / j j . LF/p ). Here we see that an increase i n the labour force p a r t i c i p a t i o n rate can be a l l absorbed by a r i s e i n the unemployment rate r e s u l t i n g i n a f a l l i n the proportion of the labour force employed with no change i n the proportion of the population employed (such as a move from \"A\" to \"B\"). I suggest that work on unemployment Insurance and the labour market unemployment has been p a r t i a l and segmented, and I b e l i e v e that a number of issues remain to be c l a r i f i e d . I f the stock of unemployed workers Is responsive to unemployment insurance, what . are the de c i s i o n s of workers employed, unemployed, and \"not i n the labour f o r c e \" that are affected? Completing the question, what are the e f f e c t s of these decisions on a l l three measured labour f o r c e states ( i . e . employment, unemployment, and \"not i n the labour f o r c e \" ) ? The meaning of measured unemployment as macroeconomic target for monetary and f i s c a l p o l i c y has been questioned by the studie s of unemployment insurance. Given that a change i n the l e v e l of unemployment or i n the unemployment rate r e f l e c t s a change i n employment, labour force p a r t i c i p a t i o n , or both these f a c t o r s , i t i s unreasonable that unemployment as a target v a r i a b l e should be set independent of e i t h e r employment or labour force p a r t i c i p a t i o n . -23-Moreover, the d i s t r i b u t i o n of the population between the labour f o r c e s t a t e s should not be the only labour market p o l i c y targets. For example, the turnover rate and the associated adjustment costs, both p r i v a t e and s o c i a l , should also be the concern of any c e n t r a l government planning. In the f o l l o w i n g chapter, I w i l l develop a model which allows an i n d i r e c t examination of worker labour force d e c i s i o n s . Although I t e x p l i c i t l y models the gross flows between labour force s t a t e s , I do not have access to gross flow data f o r the period surrounding the 1971 changes to the unemployment insurance programme. Therefore the st r u c t u r e of my model w i l l be based upon gross flows, but I w i l l estimate the numbers i n the labour force stocks ( i . e . employment, unemployment, and \"not i n the labour f o r c e \" ) . I t Is hoped that t h i s work may unravel the underlying processes that r e s u l t i n the changes to the labour f o r c e stocks. FOOTNOTES * Conrod and Kunin (1975, Mimeo) estimated the Grubel, Maki, Sax Model for nested time periods, 1953 - 1969, 1953 - 1970, 1953 - 1971, 1953-1972, and 1953 - 1973. The unemployment insurance variable, UCB/AWW (\"replacement wage\") is s t a t i s t i c a l l y insignificant at the 95% confidence level for every time period except the one used by Grubel et a l . ^ In chapter three, we present average tax rates for recipients of Unemployment Insurance benefits i n 197 3. They are approximately 10%. 3 See Green and Cousineau (1976), pp. 89-90. 4 See Grubel, Maki, and Sax (1975a), p.118. 5 The data consist of records for 2% of the Canadian Population. 6 See Swan, Steinberg, and MacRae (1976), p.24. - 25 -CHAPTER 3 - THEORETICAL APPROACH In t h i s chapter I develop a model of labour market d e c i s i o n s to e x p l a i n the flows between labour f o r c e s t a t e s . Worker d e c i s i o n s are sub j e c t to c o n s t r a i n t s imposed by the market and by the d e c i s i o n s of employers. The model i n c l u d e s the i n f l u e n c e of the unemployment insurance programme on worker's d e c i s i o n s to move i n or out of the labour f o r c e , employment, and unemployment. U n l i k e other models usi n g stocks of employment, unemployment, and \"not i n the labour f o r c e \" , t h i s model i s based upon gross flows between l a b o u r f o r c e s t a t e s . The flow process I s represented s c h e m a t i c a l l y i n the f o l l o w i n g diagram: Labour Force Flow Schematic Acceptable Job Offer^ Labour Force Entry' Employment Unemployment Not i n the Labour Force Quit o r Layo f f Labour f o r c e Withdrawal Figure 2 W i t h i n the c o n s t r a i n t s of the employer's d e c i s i o n s to l a y o f f workers o r to o f f e r employment, workers decide to q u i t , to accept job o f f e r s , to leave the labour f o r c e , o r to enter the labour f o r c e . * The worker's d e c i s i o n s are modelled as choices between a l t e r n a t i v e payoffs a s s o c i a t e d w i t h the d i f f e r e n t l abour f o r c e s t a t e s . -26-S o c i a l Roles The structure of t h i s model i s general and i s applicable to a l l age and sex groups i n the population. In other words, the processes governing labour market behaviour are conceptually the same f o r a l l members of the population. I expect, however, that the influence of the various determining factors w i l l vary according to the family role of population members. The equations of the model are generally a p p l i c a b l e , but the parameters are expected to vary between d i f f e r e n t age groups and between the sexes. For example, i n Canadian s o c i e t y men and women s t i l l assume family r o l e s which t y p i c a l l y influence t h e i r labour market behviour. It i s generally acceptable i n society that women adopt s o c i a l roles outside the labour f o r c e . It i s not generally acceptable f o r men to adopt these r o l e s . I expect t h i s s o c i a l s t r u c t u r e to cause labour market decisions made by men to be less s e n s i t i v e to small v a r i a t i o n s i n labour market incentives than decisions made by women. For men and women with f a m i l i e s , we would expect to f i n d the men committed to working and to providing the bulk of the family's wage income while women would s p e c i a l i z e i n home production of goods and services f o r the family's consumption. This view of the organization of the family and i t s members' r e l a t i o n s h i p to the labour market i s expected to hold for those age groups forming and r a i s i n g f a m i l i e s . For these age groups, I would expect that the occupations open to men and women are separated according to the employer's expected investment i n the employee. Therefore men w i l l i n general be more able to enter occupations with s i g n i f i c a n t career p o t e n t i a l than women. This suggests that men and women i n t h i s age range are probably not competing groups. The younger population which i s r e l a t i v e l y new to the labour market and i s considering f a m i l y formation would be d i f f e r e n t from the prime age group. We would expect t h i s group to be less s e t t l e d i n s p e c i f i c jobs. Also t h i s group would have l e s s established s o c i a l r o l e s . Young men with l e s s r e s p o n s i b i l i t y f o r supporting a family (a wife and/or c h i l d r e n ) are a l s o l e s s l i k e l y to be committed to the labour market than t h e i r older counterparts with these r e s p o n s i b i l i t i e s . Young women generally have established l e s s long-term and demanding s o c i a l r e s p o n s i b i l i t i e s than t h e i r older counterparts. They too may be l e s s d e s i r a b l e as career employees since there i s a r i s k that they w i l l assume t r a d i t i o n a l roles outside the labour force. Yet I expect that they w i l l p a r t i c i p a t e i n the labour market more than older women because they are not committed to roles outside the labour force or because they are i n t r a n s i t i o n between t h e i r parents' home and forming t h e i r own home. The older population l i k e the younger population generally have le s s family r e s p o n s i b i l i t i e s when t h e i r c h i l d r e n have reached the age of independence. The older male worker close to retirement i s less w i l l i n g to invest i n career advancement because the time horizon for -28-gaining a return i s too short. He therefore may become more s e n s i t i v e to short-term f l u c t u a t i o n s i n labour market incentives. Older women s i m i l a r l y released from family r e s p o n s i b i l i t i e s may substitute labour market a c t i v i t y f o r the d e c l i n i n g demands i n t h e i r conventional r o l e . These persons are u n l i k e l y to enter occupations with s i g n i f i c a n t career development requirements because of the short investment horizon and therefore are more l i k e l y to respond to short-term f l u c t u a t i o n s i n labour market incentives than t h e i r younger counterpart with more s i g n i f i c a n t family r e s p o n s i b i l i t i e s . Because the conventional family structure i s expected to influence labour market behaviour, I w i l l apply the general model to d i f f e r e n t age-sex groups. The estimated parameters are expected to vary i n accordance with the t y p i c a l l y assumed r o l e s of these age sex groups. Demographic Patterns In a d d i t i o n to conventional age-sex r o l e s , the age d i s t r i b u t i o n of the population i s expected to influence the behaviour of various groups. Probably the most i n f l u e n t i a l population d i s t r i b u t i o n i r r e g u l a r i t y f o r the labour market during the period 1966-75 i s the notorious 'baby-boom1. This post-war bulge i n b i r t h s has created havoc with each of society's i n s t i t u t i o n s that the post-war generation has entered. The 'baby-boom' has manifested i t s e l f as an excess supply of babies for adoption i n the e a r l y 1950's, followed by a dramatic increase i n the demand f o r p u b l i c education. The leading edge of the baby-boom entered the labour market i n the mid-1960's. -29-To absorb t h i s dramatic increase i n young inexperienced workers the economy has been s i g n i f i c a n t l y challenged. By estimating the general model f o r d i f f e r e n t age-sex groups, the e f f e c t s of the 'baby-boom* are i s o l a t e d p a r t i a l l y by i s o l a t i n g the 'baby-boom' age group. Undoubtedly there would be secondary displacement e f f e c t s on other age groups. S e n i o r i t y on the Job The prevalence of job s e n i o r i t y structures suggests that i t may be a f a c t o r i n a worker's attachment to a p a r t i c u l a r job. Workers who b u i l d up job s e n i o r i t y reduce the uncertainty of being employed i n the future. Because t h i s reduction i n uncertainty i s l o s t i f one leaves f o r another job, we would expect to f i n d persons with job s e n i o r i t y l e s s w i l l i n g to v o l u n t a r i l y leave t h e i r jobs than workers with l e s s job s e n i o r i t y . One f a c t o r which i s associated with job s e n i o r i t y i s age. Older workers are more l i k e l y to have est a b l i s h e d t h e i r occupation and to have b u i l t up job s e n i o r i t y than younger workers. Therefore I would expect to f i n d that younger workers are more responsive to short-term incentives such as unemployment insurance than old e r workers i n the prime labour fo r c e ages. The p a r t i t i o n i n g of the population i n t o age groups may demonstrate t h i s pattern. Segmented Labour Markets Although i t would be extreme to argue that the labour market can be l i t e r a l l y segmented, there are groups i n s o c i e t y which compete i n -30-different occupational markets. For example, the construction occupations are dominated by men while the c l e r i c a l occupations are dominated by women. To the extent that women and men are non-competing groups, cross-group effects may be isolated to age groups of the same sex category. Should a particular age-sex group's labour market behaviour be particularly sensitive to unemployment insurance, I would expect the labour market effects of the responsive group to primarily affect their sex cohorts. For example, If young women significantly adjust their labour force participation i n response to unemployment insurance, the added or reduced competition for jobs may be concentrated in markets where other women are the participants. Accounting Identities The model begins with the population identity pk \u00C2\u00BB Ek + nk + N k ( 1 ) where i s the population of a given age-sex group k in period t, and E J \u00C2\u00A3 , U ^ , are the numbers of individuals in the three labour force states (employment, unemployment, and outside the labour force respectively) that partition the age-sex group k at time t. The age groups to be examined are: 20 to 24 years, 25 to 34 years, 35 to 44 years, 45 to 54 years, and 55 years and older, and i n each age group men and women are examined separately. 2 To make i t easier to explain the model we assume at f i r s t that there Is no population change. The - 3 1 -number of Individuals i n each labour force state i n period t + 1 can then be r e l a t e d to the d i s t r i b u t i o n between stat e s In period t by the following I d e n t i t i e s (suppressing age-sex s u p e r s c r i p t s ) : E t + 1 = ( E E ) t + 1 E t + ( U E ) t + 1 U t + ( N E ) t + 1 N t (2.1) U t + 1 = ( E U ) t + 1 E t + ( ! J l T ) t + 1 U t + ( N U ) t + 1 N t (2.2) N t + 1 = ( E N ) t + 1 E t + ( U N ) t + 1 U t + ( N N ) t + 1 N t (2.3) where f o r I , j = E,U,N, ( I j ) t + i i s the proportion o f those i n s t a t e I i n period t who are i n s t a t e j i n period t+1. T h i s approach to labour market a n a l y s i s i s based on Toikka's work (1976). Although the model developed i n t h i s chapter d i f f e r s from the Toikka model i n some of the flow patterns described, the two models are of the same basic s t r u c t u r e . Toikka's model r e s t r i c t s workers who q u i t to leaving the labour f o r c e . We are i n t e r e s t e d i n persons q u i t t i n g and becoming unemployed so we have relaxed t h i s assumption. Toikka allows persons l a i d o f f to search and to f i n d employment during the same period. We have excluded t h i s two way migration between labour force s t a t e s i n order to s i m p l i f y an already complex model. The f i n a l d i f f e r e n c e i n the two models i s that Toikka r e s t r i c t s the d e c i s i o n to leave the labour force to the beginning of each period and we place t h i s d e c i s i o n at the end of each period. Although gross flow data would be p r e f e r r e d , i t i s not published f o r the period required to study the 1971 changes to the Canadian Unemployment Insurance programme. Such data w i l l be a v a i l a b l e i n the -32-early 1980's to study the 1979 changes to the Canadian Unemployment Insurance programme.3 We would hope that such a study w i l l further the understanding of the influence of unemployment Insurance on labour force decisions. Because this study w i l l use time series data, we must introduce population change to our formulation. Let (Di) t+i be the proportion of the net population change between period t and t+1 that i s i n state i in period t+1. By this we mean those who are i n state i in period t+1 but were not in the population i n period t, minus those who were in state i in period t but are not i n the population in period t+1, divided by the net population change. Let D t be the net population change (i.e. Pt+l ~ Pf.)* ^ e t n r e e identities can now be rewritten as: E t + 1 s ( E E ) t + 1 E t + ( U E ) t + 1 U t + (NE) t + 1 Nt + (DE) t + 1D t (3.1) U t + 1 \u00C2\u00A3 ( E U ) t + 1 E t + (UU) t + 1 U t + (NU) t + 1 N t + (DU) t + 1D t (3.2) N t + 1 = ( E N ) t + 1 E t + (UN) t + 1 U t + (NN) t + 1 N t'+ (DN) t + 1D t (3.3) Modelling - The General Approach Our interest i s in the determinants of the proportions ( i j ) t + i . These proportions are, however, not published, so i t w i l l be necessary to use our model to identify the effects of the unemployment insurance programme on the labour force stock data (E t, U t, N t). The model w i l l retain a structure based upon ^ the movement between states. We therefore develop below hypotheses concerning the variables and functional relations determining ( i j ) t + i and assume that. -33-( i j ) t + i = f i j (X t) + wt (4) where X i s a vector of observable variables and w i s a random disturbance term. When we su b s t i t u t e from (4) i n t o (3.1, 3.2, 3.3), we have a system of simultaneous equations with random error terms, the parameters of which we wish to estimate. The Determinants of ( i j ) , - + ] As a f i r s t step i n analysing the determinants of the flows between st a t e s , ( i j ) t . I assume that they are functions of s i x components. These include two decisions by employers and four decisions by workers. The two employer components are: F t , the p r o b a b i l i t y of an employed worker being l a i d o f f ; and R^ _, the p r o b a b i l i t y of an unem-ployed worker r e c e i v i n g a job o f f e r . The four types of worker decisions are modelled as fun c t i o n of the - payoff expectations associated with each s t a t e . The four worker components are: q t , the p r o b a b i l i t y of v o l u n t a r i l y leaving employment i n period t; R^, the p r o b a b i l i t y of accepting a job o f f e r a f t e r r e c e i v i n g i t ; d t , the p r o b a b i l i t y of withdrawing from the labour force when unemployed; and e t , the p r o b a b i l i t y of entering the labour force. The \"periods\" are months, and I assume that a l l l a y o f f s and job o f f e r s are announced at the beginning of each period, so that the p r o b a b i l i t y of q u i t t i n g during any period i s c o n d i t i o n a l upon not being l a i d o f f and the p r o b a b i l i t y of le a v i n g the labour force i f unemployed i s c o n d i t i o n a l on not rec e i v i n g an acceptable job offer. I a l s o assume that during any period each i n d i v i d u a l i n the relevant -34-labour f o r c e s t a t e makes one d e c i s i o n concerning each of the following a) Entering or l e a v i n g the labour f o r c e ( i f not employed) b) Q u i t t i n g , v o l u n t a r i l y becoming unemployed ( i f employed) c) Taking a job i f a job i s o f f e r e d ( i f unemployed) A d e c i s i o n under (a) may be combined with a d e c i s i o n under (b) or (c) i n the same period. F i n a l l y i t i s assumed that an i n d i v i d u a l who takes a job during any period does not get l a i d off or quit during the same period. Therefore an employed person may move from employment to unemployment or out of the labour force, an unemployed person may move from unemployment to e i t h e r employment o r out of the labour f o r c e , and a person \"not i n the labour force\" may move to unemployment or employment. I am therefore i m p l i c i t l y d e f i n i n g changes from one job to another job without an intervening period of unemployment as continued employment. The t r a n s i t i o n proportions, ( i j ) t , are assumed to be generated by a binomial process whose parameters are functions of these s i x p r o b a b i l i t i e s . The t r a n s i t i o n proportions i n t o unemployment are expressed as: matters (but no more than one), i f a p p l i c a b l e : ( E U ) t + 1 = [ ( 1 - F t ) q t + F t ] ( l - d t ) ( N U ) t + 1 = e t (1-R\u00C2\u00B0 Rp (5.1) (5.2) The p r o b a b i l i t y of moving from employment to unemployment depends upon l a y o f f d e c i s i o n s by firms ( F t ) , and the decisions by workers to -35-q u i t ( q t ) i f not l a i d o f f ( 1-F t) and t o remain i n the labour f o r c e ( l - d t ) . The p r o b a b i l i t i t y of e n t e r i n g the labour f o r c e and remaining unemployed depends upon the d e c i s i o n t o enter the labour f o r c e (e^) and the l i k e l i h o o d of not r e c e i v i n g an acce p t a b l e job o f f e r (1-R? The t r a n s i t i o n p r o p o r t i o n s i n t o employment are expressed as: ( U E ) t + 1 = RO RA ( 5 . 3 ) . ( N E ) t + 1 = e t R\u00C2\u00B0 R A . ( 5 . 4 ) The p r o b a b i l i t y o f an unemployed person becoming employed depends upon the p r o b a b i l i t y of r e c e i v i n g an acceptable job o f f e r (R\u00C2\u00B0 R A ) . The p r o b a b i l i t y of e n t e r i n g the la b o u r f o r c e and f i n d i n g employment w i t h i n the p e r i o d depends upon the p r o b a b i l i t y of e n t e r i n g the labour f o r c e ( e t ) and r e c e i v i n g an acceptable job o f f e r (R\u00C2\u00B0 R A).-The t r a n s i t i o n p r o p o r t i o n s out of the l a b o u r f o r c e are expressed as: ( E N ) t + 1 = [ ( 1 - F t ) q t + F t ] d t ( 5 . 5 ) ( U N ) t + 1 = ( 1-RJ R A) d t . ( 5 . 6 ) The p r o b a b i l i t y of moving from employment t o o u t s i d e the labour f o r c e depends upon l a y o f f d e c i s i o n s by f i r m s ( F t ) , and the d e c i s i o n s by workers t o q u i t ( q t ) i f not l a i d o f f ( 1-F t) and to leave the labour f o r c e ( d t ) . The p r o b a b i l i t y of l e a v i n g the labour f o r c e f o r an unemployed person i s dependent upon not r e c e i v i n g an acceptable job o f f e r (1-R\u00C2\u00B0 R A) and then d e c i d i n g to leave the labour f o r c e ( d t ) . -36-Th e t r a n s i t i o n proportions between the same labour force states are expressed as: ( E E ) t + 1 = ( l - F t ) ( l - q t ) (5.7) ( U U ) t + i = (1-RO R A ) ( l - d t ) (5.8) ( N N ) t + 1 = l - e t . (5.9) The p r o b a b i l i t y of remaining employed depends upon not being l a i d off (1-F t) and deciding not to q u i t ( l - q t ) . The p r o b a b i l i t y of remaining unemployed depends upon the l i k e l i h o o d of not re c e i v i n g an acceptable job o f f e r (1-R\u00C2\u00B0 R A) and deciding not to leave the labour force ( l - d t ) . The p r o b a b i l i t y of remaining out of the labour f o r c e depends on deciding not to enter the labour force ( l - e t ) . In pursuing the general strategy outlined here the population change v a r i a b l e s (Di) t+^ are not observable, and i n any event, are not of much i n t e r e s t i n the present context. We have no good theory concerning the determinants of these v a r i a b l e s , and so I make the s i m p l i f y i n g assumption that ( D i ) t + 1 = i t + l / P t + l (5-10) i = E,U,N Determinants of the Six Component P r o b a b i l i t i e s The f i n a l step i n developing the determinants of ( i j ) t + i i s to regard each of the s i x component p r o b a b i l i t i e s as determined by a behavioural equation. The behavioural equations are based on the evaluation by the i n d i v i d u a l of the associated payoff to d i f f e r e n t d e c i s i o n s . Although I do not develop the model of optimal strategy here, the reader i s r e f e r r e d to the job search l i t e r a t u r e (see Toikka, 1976). The following behavioural equations are based upon the assumption that i n d i v i d u a l s choose between d i f f e r e n t payoffs i n order to maximize an objective function. L a y o f f s (F) and Job O f f e r s (R\u00C2\u00B0) Some rate of l a y o f f s and job o f f e r s Is part o f \"normal turnover\". Layoffs may occur because employees are unsuitable or because of changes i n the a c t i v i t i e s of i n d i v i d u a l firms, i n the absence of changes i n general employment. Such l a y o f f s are o f f s e t by job o f f e r s . The rate of l a y o f f s i s above or below t h i s normal l e v e l when aggregate employment i s contracting or expanding. In these cases the rate of job o f f e r s i s below i t s normal l e v e l when the rate of l a y o f f s i s above i t s normal l e v e l , and v i c e versa. Thus changes i n the rate of l a y o f f s and job o f f e r s , i n opposite d i r e c t i o n s , are brought about by changes i n the rate of change of aggregate employment. Normal turnover (e.g. l a y i n g o f f employees to replace u n s a t i s f a c t o r y workers) i s more l i k e l y to occur when the labour market i s s l a c k . The rate of replacement w i l l vary with the a v a i l a b i l i t y of s u i t a b l e replacements. We assume that the unemployment rate f o r the T t o t a l population (U ) i s a proxy f o r the l i k e l i h o o d of f i n d i n g a s u i t a b l e replacement worker. The l a r g e r the stock of unemployed, the more l i k e l y i t i s that a s u i t a b l e replacement can be obtained. A change i n the desired l e v e l of employment Is assumed to r e f l e c t the expected change i n e f f e c t i v e demand (CD t). We w i l l assume that -38-businessmen predict changes i n effective demand by examining changes in actual past output. Therefore layoffs (F) and job offers (R\u00C2\u00B0) are functions of the same v a r i a b l e s , a n d CDt. These functions may be written as follows F t = f j (CDt, U*) and (6.1) - + . \" \u00E2\u0080\u00A2 RO = f 2 (CDt, Uj) . \u00E2\u0080\u00A2 . (6.2) The positive and negative symbols under each variable i n the equations are the signs of the par t i a l derivatives for the variable that are predicted by the theory. Felstein's work on temporary layoffs suggests that unemployment Insurance w i l l be used by employers and employees to make short-term adjustments in the work force. 4 He argues that the employer and employee form an 'implicit contract' to use unemployment insurance benefits to \"finance\" layoffs. This could be modelled by introducing a variable i n the layoff function (6.1) that accounts for the expected payoffs to the employee associated with continued employment and short-term unemployment. In the next section T^<-; develop the appropriate variable V t (ratio of the unemployment insurance benefit to the wage) which i s used there to measure the worker's decision to quit. If I were to model the Feldstein hypothesis e x p l i c i t l y , I would enter V t as an independent variable i n equation 6.1. I would expect that when the return to unemployment rises relative to the return to -39-employment ( V t Increasing), temporary l a y o f f s would increase. The i n t r o d u c t i o n of V t into the l a y o f f f u n c t i o n with the expected p o s i t i v e e f f e c t would not change the a n a l y s i s which follows because i t would enter the estimated equations through the term [ (1-F t) q t + F t ] (see 5.1 and 5.5). The influence of V t on both F t and q t i s expected to be p o s i t i v e . Since both F t and q t are p r o b a b i l i t i e s , r e s t r i c t e d to values between zero and one i n c l u d i n g the end points of the range, the i n f l u e n c e of V t on the term [(1-F t) q t + F t ] i s also p o s i t i v e . This means that the model developed i n t h i s study can not d i s t i n g u i s h between the F e l d s t e i n hypothesis and the hypothesis developed i n the next s e c t i o n concerning unemployment insurance-induced q u i t behaviour. Quits (q) and Job Acceptance (R A) The d e c i s i o n by a worker to qui t h i s job (or enter i n t o an i m p l i c i t l a y o f f contract of the F e l d s t e i n type) depends on the r e l a t i v e a t t r a c t i v e n e s s of remaining employed and becoming unemployed. Therefore when the payoff to unemployment r i s e s r e l a t i v e to the payoff to employment, more workers are expected to quit or enter an \" i m p l i c i t c o n t r a c t \" with the employer. The payoff to continued employment i s the current wage and the unraeasurable f u t u r e discounted b e n e f i t s of remaining employed at the present, while the payoff to unemployment includes both the personal Income obtainable while unemployed and the expected value of earnings from jobs secured by job search while unemployed. The r e l a t i v e payoff therefore i s expected to vary with the l i k e l i h o o d of obtaining unemployment insurance b e n e f i t s , the l e v e l of unemployment insurance -40-b e n e f i t s , the wage l e v e l , and the l i k e l i h o o d of obtaining a s a t i s f a c t o r y job o f f e r . The qui t f u n c t i o n can be represented as qt = f3 (CDt, uj, Vt) (6 .3) + . - +\"' where Vt i s the r a t i o of unemployment insurance b e n e f i t to the return to continued employment (the \"replacement wage\"), CDt and were previously defined as the expected change i n e f f e c t i v e demand and the unemployment rate f o r the t o t a l labour f o r c e . The value of wage o f f e r s and the number of new p o s i t i o n s are expected to vary d i r e c t l y with e f f e c t i v e demand (CDt). The higher current wage o f f e r s are r e l a t i v e to current wages, and the more job o f f e r s are being made, the greater I expect w i l l be the number of employed workers q u i t t i n g f o r job search. The greater the longer an unemployed worker would expect to search f o r acceptable employment and therefore the lower would be h i s r e s e r v a t i o n wage. To an employed worker considering q u i t t i n g the unemployment rate would be i n v e r s e l y r e l a t e d to h i s expectation of f i n d i n g a better employment s i t u a t i o n . The d e c i s i o n to accept a job o f f e r ( R A ) i s i n a sense the reverse of the q u i t d e c i s i o n . I t depends upon the r e l a t i v e expected payoffs a s s o c i a t e d with remaining unemployed and accepting a job o f f e r . Therefore job acceptance Is dependent upon the l i k e l i h o o d of obtaining a b e t t e r job o f f e r i f the worker turns down a current o f f e r . Consequently, job acceptances l i k e q u i t s depend upon the rate of good job o f f e r s and the competition f o r these job o f f e r s . The quantity and q u a l i t y (wage o f f e r ) of job o f f e r s are expected to vary d i r e c t l y with e f f e c t i v e demand (CD t). The w i l l i n g n e s s of an unemployed worker to accept a job o f f e r i s expected to be d i r e c t l y r e l a t e d to the queue of competing unemployed workers, (U^). Job acceptance i s a l s o dependent upon the expected payoff to employment (the r e a l average wage) plus the expected r e a l unemployment insurance b e n e f i t (WZ t). The expected unemployment Insurance b e n e f i t depends upon the l i k e l i h o o d that the person w i l l get a job where he w i l l be covered by unemployment Insurance. The job acceptance f u n c t i o n Is therefore w r i t t e n as R A = f 4 (CD t, U^, WZt) . (6.4) . + + Labour Force Entry (e) and Withdrawal (d) The d e c i s i o n to be i n the labour for c e depends upon the expected payoffs a s s o c i a t e d with labour force p a r t i c i p a t i o n and n o n - p a r t i c i p a t i o n . The expected return to p a r t i c i p a t i o n i n the labour f o r c e depends on the l i k e l i h o o d of re c e i v i n g an acceptable job o f f e r and the expected wage, together with the r e t u r n to unemployment (unemployment insurance b e n e f i t s ) discounted by the l i k e l i h o o d of r e c e i v i n g those b e n e f i t s . The l i k e l i h o o d of re c e i v i n g an acceptable job o f f e r depends upon the rate of job o f f e r s and the competition f o r those job o f f e r s . The -42-rate and q u a l i t y of job o f f e r s i s d i r e c t l y dependent upon CD t f o r reason given e a r l i e r . The competition f o r those jobs i s measured by the unemployment rate, UJp... The expected return to employment i s measured by the r e a l wage plus the expected unemployment insurance b e n e f i t s which depends upon the l i k e l i h o o d that the person w i l l take advantage of a job covered by unemployment insurance. The labour f o r c e entry and withdrawal decisions are seen as having the same conceptual b a s i s . Therefore, the entry and withdrawal functions may be represented as e t - f 5 (CD t, WZt) and (6.5) + - + d t = f 6 (CD t, uj, WZt) . (6.6) + -Although I have modelled the d e c i s i o n to withdraw from the labour \ \u00E2\u0080\u00A2 force as independent of whether a person q u i t s .dr i s l a i d o f f , i t Is i n t e r e s t i n g to speculate how t h i s behavior might - d i f f e r according to i . the way a person becomes unemployed. A person l a i d o f f i n v o l u n t a r i l y / would not n e c e s s a r i l y behave the same as persons on temporary l a y o f f of the F e l d s t e i n type, or persons who q u i t / t o increase job search. Persons who quit employment f o r job search would remain i n the labour f o r c e . The behavior of persons l a i d o f f may d i f f e r according to the employer's reasons f o r terminating the employee. A person on temporary l a y o f f might not engage i n a c t i v e job search ( i . e . , be out of the labour f o r c e ) . Among employees who are permanently l a i d o f f , I i . \u00E2\u0080\u00A2 \u00E2\u0080\u00A2 would expect more of those i n seasonal occupations to withdraw from -43-the labour force than those i n non-seasonal occupations. The i m p l i c a t i o n of the above i s that there may be an aggregation problem i n t r e a t i n g the labour f o r c e withdrawal Independent of the reason f o r an employee leaving employed status. The data a v a i l a b l e f o r t h i s study does not allow d i f f e r e n t i a l labour force withdrawal behaviour to be i d e n t i f i e d . In a d d i t i o n t o the s p e c i f i c a t i o n s discussed here, one could of course consider a d d i t i o n a l explanatory v a r i a b l e s to estimate labour f o r c e p a r t i c i p a t i o n behaviour. For example, Sharir and Kuch (1977) used the employment/population r a t i o f o r a l l other age-sex groups as an independent v a r i a b l e i n t h e i r labour force p a r t i c i p a t i o n equation f o r a given age group. Other v a r i a b l e s that could also be used include post-secondary school enrollment, m a r i t a l s t a t i s t i c s , and b i r t h s t a t i s t i c s , as proxy measures f o r non-labour market a c t i v i t i e s . The p o t e n t i a l of m i s - s p e c i f i c a t i o n places any e m p i r i c a l f i n d i n g s a t r i s k and consequently must be considered a l i m i t a t i o n on the r e s u l t s of i n d i v i d u a l empirical studies. However measurement and s p e c i f i c a t i o n problems are the environment of e m p i r i c a l work, and consequently i t i s always true that a f u l l e r s p e c i f i c a t i o n may a l t e r the conclusions of a study. Like the findings of e a r l i e r studies, my r e s u l t s are subject to confirmation or attack by l a t e r i n v e s t i g a t i o n . Component P r o b a b i l i t i e s Combined into the Model The twelve t r a n s i t i o n proportions i n equation system (3) can now be expressed i n terms of the exogenous v a r i a b l e s i n accordance with -44-equations (4) and (5) as follows ( E E ) t + 1 = f H (CDt, UJ, V t) + wjl (7.1) ? ? -( E U ) t + 1 = f21 (CDt, UJ, V t, WZt) + w2* (7.2) ? ? + + ( E N ) t + 1 = f31 (CDt, UT, V t, WZt) + w31 (7.3) ? ? ' + - \u00E2\u0080\u00A2 ( U E ) t + 1 = f!2 ( CD t, UT, WZt) + wj 2 (7.4) ? + + (UU) t + 1 = f 2 2 (CD t, UT, WZt) + w 2 2 (7.5) ? ? (UN) t + 1 = f 3 2 (CDt, UT, WZt) + w 3 2 (7.6) ? ? -(NE)t+l = f 1 3 (CDt, U^, WZt) + w*3 (7.7) ? ? + (NU) t + 1 = f l 3 (CDt, UJ, WZt) + w 2 3 (7.8) \u00E2\u0080\u00A2 ' ? \u00E2\u0080\u00A2 - - ? (NN) t + 1 = f 3 3 (CDt, WZt) w 3 3 ' (7.9) ( D E ) t + 1 = E t + 1 / P t + 1 + w*4 (7.10) (DU) t + 1 = U t + 1 / P t + 1 + w 2 4 (7.11) (DN) t + 1 = N t + 1 / P t + 1 + w 3 4 . (7.12) The recurrence of the explanatory variables i n the six component probabilities, i n addition to our lack of knowledge regarding fuctional forms, leave the signs of some partial derivatives without anticipated values. -45-Error terms have been added to equations (7) for two reasons. F i r s t , the actual values of ( i j ) T 4 . i thrown up by the binomial process may di f f e r from the expected values experienced in equations (5). Secondly, the influence of the exogenous variables on the six component probabilities i n equation (6) must be assumed to be subject to random error. Substituting from 7 into 3 (and simplifying the notation by suppressing the arguments of the functions fiJ^we get for (3.1) hi - fjii E t + f l J L i j t + f l 3 1 N t + ( E t + 1 / P t + 1 ) ( P t + 1 - P t) + w & Ut + wl^- N t + wljj. ( P t + 1 - P t)) which can be rewritten as E t + l / P t + l - f 1 1 E t/P t + f 1 2 U t / P t + f 1 3 N t / P t + \u00C2\u00AB ' ! \u00E2\u0080\u00A2 ' \u00E2\u0080\u00A2 z 1 t+1 t+1 t+1 t+1 where c 1 - w11 E t / P t + w 1 2 U t / P t + w 1 3 N / P + w14 Dt/Pt. t+1 t+1 C C t+1 C t+1 t+1 Rewriting and transforming 3.2 and 3.3 similarly we get the equation system Er,+1 Pt+1 \" f l l t+1 t+1 f l 3 t+1 E t/Pt \"r+1 es f21 t+1 f22 t+1 f23 t+1 \u00E2\u0080\u00A2 ut/Pt + (8) Nf+l Pt+i \u00C2\u00BB \u00E2\u0080\u0094 f 3 l t+1 f32 t+1 f3 3 t+1 Nt/Pt 3 \u00E2\u0082\u00AC t+1 _ mm where s i = wil t+1 Et/Pt + w1 2 Uf/P t+1 t+ w i 3 C t+1 Nt/Pt + w*-4 D t / P t \u00E2\u0080\u00A2 (8a) -46-U n l i k e the s t u d i e s rev iewed i n the prev ious chap te r , the model developed i n t h i s chapte r i n t e g r a t e s a d e c i s i o n sequence by workers i n t o a \" comp le te \" model o f the l abou r market. I t becomes immediate ly c l e a r t h a t the unemployment insurance programme has complex e f f e c t s on the l abou r f o r ce s t o c k s , and the r e s u l t i n g e f f e c t of unemployment Insurance on the labour f o r c e s t ock s depends upon i t s net e f f e c t upon worker d e c i s i o n s . In the f o l l o w i n g chapte r I w i l l d i s c u s s e s t i m a t i o n of the model deve loped, as w e l l as data c o n s t r u c t i o n and data sou rces . -47-FOOTNOTES * See Holt (1970) for development of this approach. Also see Denton, Feaver, and Robb (1976) for a similar approach to the Canadian Labour Market. 2 1 4 - 1 9 year olds were not Included in the study because their labour market behavior Is dominated by the educational structure. Full-time students are not e l i g i b l e for unemployment Insurance because they are not considered available for employment during the school term. 3 The revised labour force survey is designed to provide gross flow data beginning i n 1976. By 1983, 7 years of monthly data w i l l be available to examine the 1979 changes. 4 Feldstein (1976,1978) has convincingly argued that some seasonal workers and employers may use unemployment insurance during low demand periods to adjust the work force temporarily. ^ The expression (l-F)q + F i s a weighted average of q and 1 where q < 1. When both q and F r i s e , both the rise in q, one of the components of the average, and the greater weight on 1 as F rises, raises the average. - 48 -CHAPTER 4 ESTIMATION METHOD AND DATA Specification of the Model for Estimation The system of equations (8) is a \"share equation\" system. Taking into account equation (8a) we can rewrite i t as follows The share equation specification has several characteristics worthy of note. F i r s t l y , the equation system is singular. The observed shares (the elements of the column vector S t on the right hand side of (9)) must sum to unity for any t, and hence the columns of G t+i must sum to column vectors of units, and the rows of w t +j must sum to a zero vector. It has been shown that in estimating such a system one of the equations in the system i s redundant (Powell, 1969). Barten (1969) has shown that estimation of this type of equation system by the maximum likelihood method i s independent of which equation i s dropped. Secondly, the dependent variables in the share equation system are unitless and are restricted to the range [0,1]. Formulation of the Likelihood Function In order to proceed to estimation i t i s necessary to specify the functional form of the elements of the matrix G. Following the customary simplification, I assume that each element of G i s a linear s t + l \u00E2\u0080\u00A2 G t+lS t + w t + 1 S t D t/P t (9) where S t i s the column vector i^/^t (i=E\u00C2\u00BBu\u00C2\u00BBN)> Gt+i is the 3 X 3 matrix e t + i i s the 3 X 4 matrix { w^ } -49-combiriation of the exogenous v a r i a b l e s that belong to i t according t o equations (7). The elements of the matrix G are generally composed of m u l t i p l i c a t i v e component p r o b a b i l i t i e s which Implies that I n t e r a c t i v e terms should be present. The number of implied i n t e r a c t i v e terms i f the component p r o b a b i l i t i e s are assumed to be l i n e a r combinations of the exogenous v a r i a b l e s would exhaust the degrees of freedom se v e r a l times. Therefore i n t e r a c t i v e terms are excluded from the s p e c i f i c a t i o n . Dropping one equation we define the remaining two element column vec t o r of disturbance terms as (j>t with an expected value of zero and convariance matrix fit. Because the elements of (j)t are a combination of weighted random disturbance terms, I can not expect fi t to be constant over time. Given the complexity of estimation introduced when fi i s not constant, I w i l l assume fi constant. I w i l l assume th i s m i s - s p e c i f i c a t i o n of fi i s not serious and proceed. The normal density of the vector (j) i s f((J\u00C2\u00BB t) = 2 rr - 1 / 2 ( n - 1 ) . | Q | \" 1 / 2 e x p _ 1 / 2 ( $ t ) i fi-1 ^ ( 1 0 ) where n -1 i s one l e s s than the number of equations i n the system. I assumed that the vector of disturbance terms i s independently and i d e n t i c a l l y d i s t r i b u t e d between time periods. The l i k e l i h o o d function can therefore be written as f o l l o w s : L ($>) = II t f (<}),.) - T / n - 1 ) . i i _ T - 1 = 2 TT fi T exp --$-]\u00C2\u00A3 t ^ t V fi *t 0 1 ) -50-where T i s the number of time periods observed. To estimate our model by maximizing t h i s function we used the F u l l Information Maximum L i k e l i h o o d computer programme (FIML) a v a i l a b l e at the U n i v e r s i t y of B r i t i s h Columbia.' This programme maximizes the log of the l i k e l i h o o d f u n c t i o n which can be written as Woodland (1978) has pointed out that the use of a mu l t i v a r i a t e normal d i s t r i b u t i o n f o r the e r r o r vector i s a m i s - s p e c i f i c a t i o n when the range of the dependent v a r i a b l e s i s r e s t r i c t e d . He examined t h i s problem by comparing the r e s u l t s of a D i r i c h l e t d i s t r i b u t i o n (a mu l t i v a r i a t e Beta d i s t r i b u t i o n ) with a m u l t i v a r i a t e normal d i s t r i b u t i o n a p p l i e d to a share equation model. He found that the Normal model performed w e l l when the true model was a D i r i c h l e t model. Also the asymptotic l i k e l i h o o d r a t i o (chi-square) test appeared to be v a l i d f o r the m i s - s p e c i f i e d Normal model. Because the D i r i c h l e t model meets the s t o c h a s t i c s p e c i f i c a t i o n requirements of a share model and the Normal model performs well when a p p l i e d , the use of the Normal model f o r our equation system i s not unreasonable. Canadian monthly data f o r the period 1961 to 1975 i s used i n this observations are a v a i l a b l e f o r the s t a t i s t i c a l a n a l y s i s . S t a t i s t i c s Canada publishes a l l of the data s e r i e s used. l n L ( l j ) ) = - -\u00C2\u00A3(n-l) l n 2 w -_T l n _ ( $ T ) . Q -1 (J)T. (12) Data study. One hundred and eighty monthly seasonally unadjusted The labour force data are collected from the Labour Force Survey (Statistics Canada, 71-001). The questionnaire for the survey was conceptually revised i n 1976. The new method does not permit linking the two series (except i n an arbitrary fashion) and Statistics Canada has only produced the revised series back to 197 0 for selected groups. Consequently we use the old series, which contains more Information (observations) for our study. This unfortunately requires that we stop at December 1975. Our observations start in 1961 because age-sex data f i r s t became available i n January 1961. Two data series for the total labour force are collected: the total number of unemployed, UL^; and the total number of employed, E^. The unemployment rate for the total labour force is calculated (by Statistics Canada) by the following formula UT= ULj / (Ej + ULj) . (13) o In our notation the superscript T stands for \"t o t a l \" , i n contrast to particular age-sex groups. The variable i n our equation for the expected change In effective demand (CDt) i s not observable. The businessman presumably uses a process of observing past trends i n output, outstanding orders, inventories, and other factors to determine his labour requirements in the short run. We w i l l use the change in the real output index as a proxy for CDt. The real industrial product (RIP t) is available from Statistics Canada (61-005) for a l l Canadian industries (standard i n d u s t r i a l c l a s s i f i c a t i o n ) . This s e r i e s i s chosen i n preference t o Gross National Expenditure (GNE) data, p r i m a r i l y because I t i s a v a i l a b l e monthly and GNE i s a q u a r t e r l y s e r i e s . We formulate the proxy f o r CD t as CD t = RIP t_! - RIP t_i3 (14) ( i . e . , the year-to-year change i n output, lagged one month). The Canadian consumer pr i c e index (CPI t) i s a v a i l a b l e from S t a t i s t i c s Canada (62-010). We use the CPI f o r a l l items to convert the average weekly wage and the average unemployment insurance b e n e f i t to r e a l terms. There are t h e o r e t i c a l reasons to believe that the monetary wage does not Include a l l the f a c t o r s that compose the cost of leaving a job. Career p o t e n t i a l , p o t e n t i a l future earnings, job s e c u r i t y , f r i n g e b e n e f i t s , and non-pecuniary be n e f i t s are among the other f a c t o r s which would a f f e c t a worker's evaluation of a job. In the absence of a more i d e a l measure, the average weekly wage (AWWt) published by S t a t i s t i c s Canada (72-002) i s used as a proxy f o r the worker's evaluation of the job. The average weekly wage Is adjusted f o r i n f l a t i o n by CPI as follows Wt = AWWt / CPI t . (15) The three unemployment insurance s t a t i s t i c s c o l l e c t e d a r e : the population covered by the programme ( U I P t ) , the number of unemployment insurance b e n e f i c i a r i e s (UIR t), and the average gross unemployment insurance b e n e f i t (GUIB t). These programme s t a t i s t i c s are a v a i l a b l e -53-in Statistics Canada publication, The Statistics on the Operation of the Unemployment Insurance Act (73-001). In 1972, the average gross unemployment insurance benefit became taxable for the f i r s t time. We adjust the gross unemployment insurance benefit received after 1971 by the average tax rate for the age-sex populations which received unemployment insurance benefits in 1973. These tax rates were calculated using the Consumer Finance Survey micro data base for 1973. These average tax rates approximate 10% which i s one of the levels assumed by Grubel, Maki, and Sax (1975a) i n their reply to Kaliski (1975). The adjustment tax factors are as follows: (1 \u00E2\u0080\u0094 average tax rate) Men Women 20-24 Years 0.9091 0.9518 25-34 Years 0.8996 0.9411 35-44 Years 0.9059 '. 0.9582 45-54 Years 0.9028 0.9145 55-64 Years 0.9098 0.9350 The average unemployment insurance benefits are adjusted as follows UIBi = GUIBt (l-TX 1) for t > 1972 t (16) UIBj \u00C2\u00BB GUIBt for t < 1972 where 1 indicates the age-sex group and TX 1 is the average tax rate for the i t n age-sex group in 1973. -54-v i i s a measure of the expected r e l a t i v e payoff f o r unemployment and employment relevant to the de c i s i o n to q u i t . We ca l c u l a t e t h i s v a r i a b l e as the r a t i o of the average unemployment insurance benefit (UIB^) net of tax to the average weekly wage (AWWt) , discounted by the proportion of the employed population covered by the unemployment insurance programme. V\u00C2\u00A3 i s thus c a l c u l a t e d as V i = [(UIP t - UIR t)/ET] [uiB^/AWWt-] . (17) Z\u00C2\u00A3 i s a measure of the l i k e l i h o o d of being covered by the unemployment insurance programme m u l t i p l i e d by the r e a l average unemployment insurance b e n e f i t . This v a r i a b l e i s ca l c u l a t e d as Z i = [(UIP t - UIR t)/FT] [UIB^/CPI,-] . (18) Because coverage by unemployment insurance was expanded to v i r t u a l l y a l l employees i n 1972, (UIP t - UIR t)/E T t increased i n 1972. The maximum unemployment insurance b e n e f i t s were increased i n 197 2, consequently average unemployment insurance b e n e f i t s also increased. WZ\u00C2\u00A3 i s the composite v a r i a b l e f o r the expected return from employment (W t) and the expected return from unemployment insurance i f a job i s taken ( Z j ) . The var i a b l e i s cal c u l a t e d as follows WZ1 = Wt + z j t z t (19) -55-The composite v a r i a b l e WZ was constructed f o r the e m p i r i c a l a n a l y s i s because i t s components were expected to have e f f e c t s of the same nature ( s i g n ) , and because the composite v a r i a b l e saved degrees of freedom and reduced c o l l i n e a r i t y between the exongeneous v a r i a b l e s . The degrees of freedom were of more concern e a r l y i n the research when i n t e r a c t i v e v a r i a b l e s were being considered. The c o r r e l a t i o n c o e f f i c i e n t between W and Z i s 0.87. The s e n s i t i v i t y of the r e s u l t s to the method of constructing the composite v a r i a b l e , WZ, was examined. The r e l a t i v e weights on the components of WZ were set at 100%-0%, 70%-30%, 50%-50%, 30%-70%, and 0%-100%. In a l l cases c o e f f i c i e n t s which changed signs between weighting experiments e x h i b i t e d very low t-values. No s t a t i s t i c a l l y s i g n i f i c a n t c o e f f i c i e n t changed signs. Therefore the a n a l y s i s and q u a l i t a t i v e conclusions about the influence of other v a r i a b l e s are not s e n s i t i v e to the weights assigned to the components of WZ. The c o e f f i c i e n t s f o r the v a r i a b l e WZ by themselves do not n e c e s s a r i l y describe the i n f l u e n c e of the separate components, since the components W and Z are h i g h l y c o r r e l a t e d . CHAPTER 5 RESULTS In chapter three the t h e o r e t i c a l basis f o r an economic determination of the movements by people between labour f o r c e states was developed. While t h i s Is the focus of our study, I t Is conceivable that these movements during the observed period (1961-1975) were unresponsive to economic f a c t o r s . This case i s the n u l l hypothesis i n a test to determine whether the economic model i s s t a t i s t i c a l l y s i g n i f i c a n t . Before the economic model i s developed and tested, the case of constant t r a n s i t i o n p r o b a b i l i t i e s w i l l be examined which r e f l e c t s t h i s n u l l hypothesis. E m p i r i c a l A n a l y s i s - Assuming Constant T r a n s i t i o n P r o b a b i l i t i e s The system of equations (8) (reproduced below) can be taken to describe the case of constant t r a n s i t i o n p r o b a b i l i t i e s when a l l the elements are regarded as constant c o e f f i c i e n t s f o r the lagged labour force state shares. E t + l / p t + l f l l f l 2 f l 3 E t/P t \" e t i l \" u t+i/ pt+i = f21 f22 f23 \u00E2\u0080\u00A2 u t/ pt + N t + l / p t + l f31 f32 f33 N t/P t \u00C2\u00BB -The estimated t r a n s i t i o n p r o b a b i l i t y matrices f o r the ten age-sex groups are reported In Table 1. Estimated elements of the matrices which are i n s i g n i f i c a n t at the 5% l e v e l are reported i n parentheses. The i n s i g n i f i c a n t estimated t r a n s i t i o n p r o b a b i l i t i e s occur i n a l l the male populations and i n the oldest female population. There are no -57-estimated elements that exceed the theoretical l i m i t of one. Negative estimated elements exist In each population, and are significant in 7 of the 10 cases. The significant negative elements are either f ^ 3 o r f 3 2 or both which are the transition probabilities between unemployment and \"outside the labour force\". The signs of these estimated elements are as follows Table 1 Estimated Constant Transition Probabilities 3 E i = i f l l f l 2 f13 EE UE NE f21 f22 f23 as EU UU NU f 3 * f32 f33 EN UN NN f i j - 1.0 for j - 1,2,3 20 - 24 Years Old .9400 ,0161 .0439 Women .3952 .8651 -.2603 .0525 .0112 .9587 .9561 (-.0003) .0436 Men .2997 .8182 (-.1179) (.0775) .1041 .8184 25-34 Years Old Women .9592 .0615 .0139 .0161 . 77 28 -. 0047 .0247 .1657 .9908 .9952 (.0001) .0047 Men -.2515 .7838 -.0353 (-.2478) .3316 .9162 -58-Table 1 Estimated Constant Transition Probabilities (Cont.) 35-44 Years Old .9514 .0173 .0313 Women .8129 .7051 -.5180 .0183 .0054 .9871 .9985 (-.0024) .0039 Men .2459 .7764 (-.0223) (-.3363) .4464 .8899 45-54 Years Old ,9564 ,0170 0266 Women .4400 .6786 -.1186 .0214 ,0056 ,9842 .9935 (.0018) .0047 Men .1996 .8353 .0349 (-.0465) .1036 .9429 55 Years and Older .9378 (.0097) .0525 Women (.0960) .6314 (.2726) .0117 (-.0002) .9885 .9803 (.0045) (.0152) Men .3494 .7499 ,0993 (.0007) (.0101) .9892 -59-MEN WOMEN 20-24 25-34 35-44 45-54 55 Years 20-24 25-34 35-44 45-54 55 Years Years Years Years Years and Over Years Years Years Years and Over f23 + - - - - + + +^ (+) (+) f32 \u00E2\u0080\u0094 + \u00E2\u0080\u0094 \u00E2\u0080\u0094 (+) (-) \u00E2\u0080\u0094 (-) \u00E2\u0080\u0094 \u00E2\u0080\u0094 These negative estimates are inconsistent with the theory which predicts values i n the range [0,1] Inclusive. They therefore suggest that the null hypothesis i s not valid. Dynamic Properties of the \"Null Hypothesis Model\" A property of the model i s Its a b i l i t y to cycle. This can be shown by examining the complementary function which depends upon the elements of transition matrix. The general form of the model is as follows: \" a i l a 1 2 a13~ \" E t - l / P t - f \" ' i t \" ut/Pt = 321 322 a23 \u00E2\u0080\u00A2 \"t-l/Pt-1 + *2t Nt/Pt_ -a31 a32 a33 . -Nt-l/Pt-1- - e 3 t -Because two equations determine the system they also determine i t s dynamic properties, therefore we partition the transition matrix as follows: _ A = a l l a12 ! a13 a21 a22 I a23 _a31 a32 ' a33 L l l A 31 J A 1 3 i i_ l a33 -60-and the vector as follows Et/Pt\" \" * t ' ' i t \" \u00E2\u0080\u00A2 t ut/Pt \u00E2\u0080\u00A2 = and *2t N t/Pt N t/P t - \u00E2\u0082\u00AC 3 t - - C 3 t . The equation can be rewritten as A U A 1 3 \u00E2\u0080\u00A2 Nt-l/Pt-1 + Nt/PtJ = A 3 1-a 3 3 e 3 t where the systems a d d i t i v e c o n s t r a i n t s s t i l l hold pfx,- 1= 1 , l'A = 1\u00C2\u00BB, and l f ^ t l = 0 [ N t / P t J f 3 t j when 1' = (1, 1, 1). I w i l l examine the dynamic properties of the system X t = A U X t_x + A 1 3 ( N t _ i / P t _ i ) + t S u b s t i t u t i n g i n t o the above Nt - l/Pt-1 = 1 - 1 ' X t _ i derived from the a d d i t i v e constraint we have X t = ( A H - A 1 3 1') X t_! + A 1 3 + 23 t+1 t+1 t+1 f ? l f??, f?3 S D t + l \u00E2\u0080\u009E1/Pl + S D t + l + N'/pi S D c + l c 3.1 6 t+1 where SDJ^ i s a seasonal c o e f f i c i e n t that has a d i f f e r e n t estimated value f o r eleven months of the year and i s r e s t r i c t e d to zero f o r the twelfth. The seasonal c o e f f i c i e n t s are estimated by means of dummy va r i a b l e s f o r the eleven months, the dummy v a r i a b l e f o r a given month takes the value of one during that month and zero f o r every other month. The seasonal c o e f f i c i e n t s f o r the three equations are -70-restricted to sum to zero for any period. /. SDt = 0, for any value o f t k=i thus preserving the share equation specification i n chapter 4. When seasonal correction i s included, the interpretation of the coefficients of the economic variables i s that they represent the influence of the economic variables on the seasonally adjusted dependent variables. Other specifications of seasonal adjustment are possible but they tend to exhaust the degrees of freedom available i n the data. For example, monthly seasonal coefficients could be introduced for each of the economic variables by multiplying the economic variables by seasonal dummy vector for each month. This approach would consume an additional two hundred andJ ninejdegrees of freedom which are not available i n the data range. Because I am not primarily interested i n the seasonal pattern of the labour force stocks, I have not concentrated the information available i n the data on this subject. The maximum likelihood estimates for the coefficients of the exogeneous variables and lagged dependent variables when estimated with seasonal correction are presented i n Tables 3.1, 3.2, 3.3, 3.4, 3.5, and 3.6. The coefficients for the seasonal dummies are not presented here. Note that the Durbin-Watson S t a t i s t i c s for a l l the equations allow us to reject the hypothesis that autocorrelation in Table 3.1 Estimated Economic Model with Seasonal Dummy Var i a b l e s Dependent V a r i a b l e : Et+i/?t+\ Men 20-24 Years 25-34 Years 35-44 Years 45-54 Years 55 Years + Older E t / P t 0. 9773 (63.34) 0.9874 (78. 54) 0.9785 (91.31) 0.9878 (72. 53) 0.8024 6 (10.73) (CD t.E t)/? t -0. 1048 E-3 (0.89) -0.6731 E-4 (0.7 0) -0.3689 E-4 (0.51) -0.1176 E-3 (1.46) -0.1441 E-3 (0.40) (uT.E t)/P t -0.9908 (3.55) -0.6496 (3.16) -0.6330 (3.53) -0.6518 (3.09) 1. 07 53 (0.92) ( V t . E t ) / P t -0.01428 (0.42) 0.01521 (0.62) 0.01381 (1.10) -0.02626 (1.60) -0.05855 (1.64) U t/P t 0.7801 (3.16) 0.3571 (0.79) 0.5443 (1.53) 0.5148 (1.31) -1.04831 (1.43) ( C D t . U t ) / P t -0.7071 E-3 (0.69) 0.9860 E-4 (0.06) 0.1211 E-2 (1.00) -0.1259 E-3 (0.10) 0.3284 E-2 (1.33) (uT.U t)/P t -0.4009 (0.31) -0.5613 (0.30) 0.6998 (0.47) 0.3910 (0.24) 5.3536 (1.81) (WZt.Ut)/Pt -0.2391 E-2 (1.98) 0.1698 E-3 (0.07) 0.3664 E-3 (0.19) -0.9428 E-3 (0.42) 0.6895 E-2 (1.55) Nt/Pt -0.2250 (1.68) -0.1541 (0.16) -0.2928 (0.32) -0.3202 (0.51) 0.3353 (2.52) (CDt.Nt)/Pt 0.1371 E-2 (2.31) 0.2205 E-2 (0.65) 0.5376 E-3 (0.14) 0.3007 E-2 (1.59) 0.1228 E-4 (0.03) (uT.N t)/P c 0.2231 (0.16) 10.034 (1.53) -1.8195 (0.27) 4.9480 (1.18) -2.1209 (1.45) (WZt.Nt)/Pt 0.1211 E-2 (1.39) -0.2132 E-2 (0.37) 0.3190 E-2 (0.68) 0.1738 E-2 (0.62) -0.6418 E-3 (1.76) R2 = 0.98 R2 = 0.92 R 2 = 0.96 R2 = 0.96 R2 = 0.94 O.W.- 2.24 D.W.= 1.76 D. W.= 1.80 D.W.= 1.61 D. W.= 1.83 -72-Table 3.2 Estimated Economic Model with Seasonal Dummy Variables Dependent V a r i a b l e : U t+i/P t- r.i Men 20-24 Years 25-34 Years 35-4 4 Years 45-54 Years 55 Years + Older E t/P t 0.05534 (3.05) 0.02259 (1.76) 0.01996 (1.88) 0.01365 (0.95) 0.06837 (0.62) (CD t.E t)/P t 0.7629 E-4 (0.74) 0.4826 E-4 (0.52) -0.1525 E-4 (0.25) 0. 277 6 E-4 (0.44) 0. 1256 E-3 (0.39) (UT.E t)/P t 1.1044 (4.44) 0.5862 (2.88) 0.5259 (3.24) 0.5873 (3.33) 1.0300 (0.95) (V t.E t)/P t 0.09795 (2.99) 0.01394 (0.58) 0.3372 E-2 (0.31) 0.01454 (1.10) 0.01485 (0.42) (WZt.Et)/Pt -0.4094 E-3 (2.31) -0.1633 E-3 (2.99) -0.5631 E-4 (1.04) -0.1056 E-3 (1.16) -0.5082 E-3 (0.80) Ut/Pt -0.07800 (0.33) 0.5337 (1.22) 0.5965 (1.95) 0.3543 (1.13) 0.1145 (0.18) K X j ,... ,X n;9i ,02,... .\u00C2\u00A9k) where Xj,....,X n are the observed sample values, 9j,02>\u00E2\u0080\u00A2\u00E2\u0080\u00A2\u00E2\u0080\u00A2,9^ are estimated c o e f f i c i e n t s and 0^, Q,^ ,. \u00E2\u0080\u00A2 \u00E2\u0080\u00A216^ c o e f f i c i e n t s that have p a r t i c u l a r values according to the hypothesis being tested. When the sample s i z e i s la r g e , the random v a r i a b l e , -2L nX has approximately the X 2 d i s t r i b u t i o n with h degrees of freedom (given the 9^). i f we set s e l e c t e d c o e f f i c i e n t s equal to zero, we can t e s t whether that can s t a t i s t i c a l l y be accepted as t h e i r true value, thus t e s t i n g a h u l l hypothesis. I w i l l t e s t whether the seasonal dummy v a r i a b l e c o e f f i c i e n t s , the unemployment insurance v a r i a b l e c o e f f i c i e n t s (V, and WZ), the other economic v a r i a b l e c o e f f i c i e n t s (CD, and U^), and combinations of these c o e f f i c i e n t s can be s t a t i s t i c a l l y considered equal to zero. Tables 4.1 and 4.2 present the Ln of the maximized l i k e l i h o o d functions requried f o r our t e s t s . The t e s t r e s u l t s are presented In tables 5.1 and 5.2 f o r men and women. -78-Table A. 1 Ln of the L i k e l i h o o d Functions Economic Model w / S e a s o n a l C o r r e c t i o n MEN 20-24 Y e a r s 25-34 Ye a r s 35-44 Y e a r s 45-54 Ye a r s 55 Y e a r s + Ol d e r F u l l Model w / CD, UT,V,WZ .Seasonal Dummies, and Co n s t a n t s 1232.22 1502.94 1659.65 1584.48 1406.58 Model W/CD,U T, S e a s o n a l Dummies and C o n s t a n t s 1221.52 1486.12 1629.18 1568.62 1392.75 Model W/V, WZ, S e a s o n a l Dummies and Co n s t a n t s 1199.86 1488.94 1630.61 1569.79 1386.30 Model w / S e a s o n a l Dummies, and C o n s t a n t s 1186.88 1473.88 1613.63 1557.02 1371.61 Economic Model Q/o S e a s o n a l C o r r e c t i o n 20-24 Y e a r s 25-34 Y e a r s 35-44 Y e a r s 45-54 Ye a r s 55 Years + O l d e r F u l l Model w / CD, U T,V,WZ,and C o n s t a n t s 917.133 1339. 15 1494.82 1425.99 1311.42 Model w / CD,U T, and C o n s t a n t s 901.529 1313.87 1468.00 1406.45 1291.11 Model w / V, WZ, and C o n s t a n t s 891.840 1324.69 1482.69 1417.60 1295.66 Model w / C o n s t a n t s 883.488 1308.59 1463.70 1399.49 1278.90 - 7 9 -Table 4.2 Ln of the L i k e l i h o o d Functions Economic Model w / Seasonal C o r r e c t i o n WOMEN 20-24 Years 25-34 Years 35-44 Years 45-54 Years 55 Years + Older F u l l Model w / CD, UT,V,WZ,Seasonal Dummies, and Constants 1401.34 1621.95 1610.57 1578.23 1743.53 Model W/CD,U T, Seasonal Dummies and Constants 1376.21 1559.03 1593.83 1566.94 1729.03 Model W/V, WZ, Seasonal Dummies and Constants 1375.83 1603.09 1587.65 1561.96 1724.11 Model w / Seasonal Dummies, and Constants 1355.78 1584.56 1565.69 1537.58 1703.92 Economic Model w/o Seasonal C o r r e c t i o n 20-24 Years 25-34 Years 35-44 Years 45-54 Years 55 Years + Older F u l l Model w / CD, UT,V,WZ,and Constants. 1305.56 1503.29 1501.26 1471.88 1680. 59 Model w / CD,UT, and Constants 1283.16 1484.00 1487.79 1454.79 1668.15 Model w / V, WZ, and Constants 1292.74 1489.63 1484.18 1455.93 1668.18 Model w / Constants 1273.05 1474.74 1473.06 1438.67 1651.84 -80-Table 5.1 Li k e l i h o o d Ratio Tests F u l l Model w / CD,UT,V,WZ, Seasonal Dummies and Constants Likelihood Ratio Test Degrees of Freedom MEN 20-24 Years 25-34 Years 35-44 Years 45-54 Years 55 Years + Older Test: Seasonal Dummies 22 630.17 327.58 329.66 316.98 1Q0.32 Test: V, and WZ 7 21.40 33.64 60.94 31.72 27.66 Test: CD, and UT 12 64.72 * 28.00 58.08 29.38 40.56 Test: CD,UT, V, and WZ 19 90.68 58.12 92.04 54.92 69. 94 Test:CD,U T, V.WZ, and Seasonal Dummies 41 697.48 388.70 391.90 369.98 255. 36 F u l l Model w / CD,UT,V, WZ, and Constants 20-24 Years 25-34 Years 35-44 Years 4 5-54 Years 55 Years + Older Test: V, and WZ 7 31.21 50. 56 53.64 39.08 40.62 Test: CD, and UT 12 50.59 28.92 ** 24.26 *** 16.78 31.52 Test: CD.liT and WZ 19 67.29 62.12 62.24 53.00 65.04 Note: A l l tests are s i g n i f i c a n t at 99.5% l e v e l except, * 99% l e v e l , ** 97.5% l e v e l , and *** 85% l e v e l . -81-Table 5.2 L i k e l i h o o d Ratio Tests F u l l Model w / CD,UT,V,WZ, S e a s o n a l Dummies and C o n s t a n t s L i k e l i h o o d R a t i o T e s t Degrees o f Freedom WOMEN 20-24 Y e a r s 25-34 Ye a r s 35-44 Years 45-54 Ye a r s 55 Y e a r s + Olde r T e s t : S e a s o n a l Dummies 22 191.56 237,^2 218.62 212.70 12S.88 T e s t : V, and WZ 7 50.26 125.84 33.48 22.58 29.0 T e s t : CD, and UT 12 51.02 37.72 45.84 32.54 38.84 T e s t : CD,u T, V, and WZ 19 91.12 74.78 89.76 81.30 79.22 T e s t : C D , U T , V.WZ, and Sea s o n a l Dummies 41 256.58 294.42 275.02 279. 12 183.38 F u l l Model w / CD,u T,V, WZ, and Co n s t a n t s 20-24 Y e a r s 25-34 Ye a r s 35-44 Years 45-54 Y e a r s 55 Y e a r s + O l d e r T e s t : V, and WZ 7 44.80 38. 58 26.94 34.18 24.88 T e s t : CD, and UT 12 ** 25.64 * 27.32 34.16 31.90 ** 24.32 T e s t : CD,U T and WZ 19 65.02 57. 10 56.40 66.42 57.50 Note: A l l t e s t s a r e s i g n i f i c a n t a t 99.5% l e v e l e x c e p t , * 99% l e v e l , and ** 97.5% l e v e l . -82-The hypothesis that the seasonal dummy coefficients are zero can be rejected at the 99.5% level for a l l ten equation systems. The hypothesis that the unemployment insurance coefficients are zero can be rejected at the 99.5% level for a l l ten equation systems whether seasonal dummy variables are specified or not. When seasonal dummy variables are specified with the economic model, the hypothesis that the economic variable (CD, and U^) coefficients are zero can be rejected at the 99.5% level 9 times and at the 99% level for the remaining coefficient. When seasonal dummy variables are not specified with the economic model, the hypothesis that the economic variable (CD, and U^) coefficients are zero can be rejected at the 99.5% level 5 time,at the 99% level once, at the 97.5% level 3 times, and at the 85% level for the last one. These results justify the inclusion of the explanatory variables in the model according to the theoretical discussion i n Chapter 3. Only i n one case (men 45-54 years old) does a test result f a l l below a 97.5% lev e l . Because these variables are highly significant for the other age-sex groups, we consider their inclusion in the model for a l l age-sex groups to be warranted. Analysis of the Economic Moder The addition of seasonal correction greatly improves the R-Square and Durbin-Watson s t a t i s t i c s for men. It also changes many regression coefficients rendering many non-significant. These results are summarized in Table 6 for those coefficients for which our theory predicts the sign. -83-TABLE 6 Summary of Economic Model Estimates A G E - S E X GROUP H E N 20-24 Years 25-34 Years 35-44 Years 45-54 Years 55 Years + Older R 2 D.W. Equation 1 E/P Dependent Variable 0.674 1.27 0.981 2.24 0.770 1.10 0. 924 1.76 0.780 0.850 0.963 1.80 0.769 0.88 0.956 1.61 0.863 1.26 0.944 1.83 R 2 D.H. Equation 2 U/P Dependent Variable 0.850 1. 23 0.930 1.57 0.786 1.11 0.900 1.69 0.801 0.815 0.961 1.69 0. 793 0.78 0.958 1.64 0.628 1.44 0.767 1.71 R 2 D.W. Equation 3 N/P Dependent Variable 0.609 1.44 0.971 2.21 0.773 1.63 0. 942 1.99 0.704 1.49 0.868 2.11 0.836 1.61 0.914 2.01 0.946 1.52 0.970 2.01 SIGNS OF COEFFICIENTS AND t VALUES Effect of V(-) on E\u00E2\u0080\u0094\u00C2\u00BB-E - 2.4 (~)0.4 - 2.3 (+)0. 6 - 3.0 (+)1.1 - 2.3 (-)1.6 (-)0.9 C-)1.6 Effect of V(+) on E \u00E2\u0080\u0094 \u00C2\u00BB U + 4.4 + 3.0 + 3.1 (+)0.6 + 3.5 (+)0.3 (+)1.9 (+)1.1 (+)0.3 (+)0.4 Effect of V(+) on E\u00E2\u0080\u0094\u00C2\u00BB>N (+)1.1 - 2.3 (- )0.7 - 4.1 (+)0.6 - 3.1 + 2.3 (+)1.3 (+)0.9 (+)1.5 Effect of WZ(+) on E-*\u00C2\u00BBU - 3.3 - 2.3 - 3.3 - 3.0 (~)1.2 (-)i . o (~)1.6 (-)1.2 (+)0.0 (-)0.8 Effect of WZ(-) on E\u00E2\u0080\u0094\u00C2\u00BB>N + 3.3 + 2.3 + 3.3 + 3.0 (+)1.2 (+)1.0 (+)1.6 (+)1.2 (-)O.O (+)0.8 Note: F i r s t row estimates are seasonally uncorrected, second row estimates are seasonally c o r r e c t e d . -84-TABLE 6 Con't SIGNS OF COEFFICIENTS and t VALUES AGE-SEX MEN GROUP 20-24 Ye a r s 25-34 Years 35-44 Years 45-54 Ye a r s 55 Y e a r s + O l d e r E f f e c t o f (-)O.l (+)0.3 (~)0.3 C-)0.5 (+)1.8 U T(+) on U \u00E2\u0080\u0094 \u00C2\u00BB E <->0.3 (-)0.3 (+)0.5 (+)0.2 (+)1.8 E f f e c t o f (+)0.6 (-)0.2 (-)o.o (+)0.8 (+)0.3 UT(-) on U \u00E2\u0080\u0094 U (+)1.4 (+)0.3 (-)1.4 (-)O.l (+)0.7 E f f e c t o f (+)1.7 (+)1.9 (-)0.9 (-)1.7 (+)1.6 WZ(+) on U\u00E2\u0080\u0094\u00C2\u00BB-E + 2.0 (+)0.1 (+)0.2 (-)0.4 (+)1.6 E f f e c t of - 2.0 - 2.3 (+)1.8 (+)0.5 - 2.8 WZ(-) on U-*-N (+)0.7 <-)0.5 (+)0.9 (-)0.6 - 2.8 E f f e c t o (+)0. 0 (-)0.2 ( - ) l . l (~)0.5 (+)0.6 CD(-) on N.-^K (->1.6 (-)O.l (-)1.9 - 2 . 5 (+)0.4 E f f e c t of < ; (+>0.1 (+)0.2 (-)0.2 (~)1.5 (+)0.0 u T ( - ) on N-*U + 2.4 (-)0.9 (+)1.3 ( - ) l . l (\")0.6 E f f e c t of + 2.0 (+)1.3 (~)1.6 (-)0.5 + 2.2 UT(+) on N\u00E2\u0080\u0094-N - 2.3 (-)1.7 (-)1.7 (-)0.3 + 2.6 E f f e c t of <-)0.8 (+)0.2 + 2.3 + 2. 1 (~)0.8 WZ(+) on N\u00E2\u0080\u0094\u00C2\u00BB-E (+)1.4 (-)0.4 (+)0.7 (+)0.6 (\")1.3 E f f e c t of (+)0.4 (+)0.3 (-)1.4 ( - ) l . l (+)1.3 WZ(-) on N\u00E2\u0080\u0094*N (+)1.2 (+)0.9 (+)0.2 (+)0.3 (+)0.4 Note: F i r s t row estimates are seasonally uncorrected, second row estimates are seasonally c o r r e c t e d . -85-TABLE 6 Con't AVERAGE ELASTICITIES AGE-SEX GROUP MEN 20-24 Years 25-34 Years 35-44 Y e a r s 45-54 Y e a r s 55 Y e a r s + O l d e r 1 Change In E/P -0.0019 -0.007 2 0.0469 -0.0062 0.0446 0.0179 0.0328 0.0017 0.0208 -0.0394 A b s o l u t e Change i n E/P -0.0000149 -0.0000564 0.0004265 -0.0000561 0.0004148 0.0001667 0.0002988 0.0000155 0.0001081 -0.0002056 % Change i n U/P -0.2159 -0.2215 -1.1233 -0.3033 -1.0158 -0.6546 -0.9835 -0.4203 -1.3024 -0.3817 A b s o l u t e Change i n U/P -0.0001702 -0.0001746 -0.0005502 -0.0001469 -0.0004573 -0.0002852 -0.0004014 -0.0001794 -0.0003659 -0.000107 9 % Change i n N/P 0.1343 0.1682 0.3656 0.4392 0.1713 0.4887 0.2137 0.3543 0.0571 0.0697 A b s o l u t e Change i n N/P 0.0001387 0.0002314 0.0001120 0.0002034 0.0000428 0.0001179 0.0001028 0.0001635 0.0002583 0.0003137 % Change i n U/(E-HJ) -0.1945 -0.1948 -1.1119 -0.2821 ' -1.0114 -0.6426 -0.9728 -0.4032 -1.2559 -0.3247 A b s o l u t e Change i n U/(E+U) -0.0001778 -0.0001781 -0.0005556 -0.0001433 -0.0004670 -0.0002369 -0.0004170 -0.0001804 -0.0005444 -0.0001671 Note: F i r s t row estimates are seasonally uncorrected, second row estimates are seasonally c o r r e c t e d . -86-TABLE 6 Con't AGE-SEX GROUP WOMEN 20-24 Years 25-34 Years 35-44 Ye a r s 45-54 Years 55 Y e a r s + O l d e r R 2 D.W. E q u a t i o n 1 E/P Dependent Value 0.930 1.87 0. 963 2. 11 0.986 1.77 0.994 2.24 0.974 1.79 0.988 2.25 0.947 1.78 0.976 2.07 0.884 2.00 0.931 2.14 R 2 D.W. E q u a t i o n 2 U/P Dependent V a r i a b l e 0.908 2.19 0.942 2.04 0.922 1.92 0.954 1.97 0.882 1.95 0.925 2.10 0.710 1.93 0.807 2.13 0.545 1.99 0.623 2.08 R 2 D.W. E q u a t i o n 3 N/P Dependent V a r i a b l e 0.954 1.87 0.974 21.0 0.987 1.98 0.995 2.23 0.978 1.79 0.989 2.21 0.954 1.82 0. 978 2.05 0.880 1.94 0.928 2.05 SIGNS OF COEFFICIENTS AND t VALUES E f f e c t o f v(-) on E\u00E2\u0080\u0094*-E - 2.9 - 2.9 - 2 . 7 - 2 . 9 - 2.0 (-)1.7 - 2.7 (~)1.7 - 3.1 - 3.6 E f f e c t o f V(+> on E - \u00C2\u00BB U (+)0.0 (-)0.3 (+)1.7 (+)0.7 (-)0.5 - 3.3 (- )0.0 (- )0. 9 ( + ) i . o (+)1.8 E f f e c t o f V(+) on E-*-N + 2.9 + 2.9 + 2.0 + 2.4 + 2.2 + 2.9 + 2.6 + 2.0 + 2.7 + 2.6 E f f e c t of WZ(+) on E-\u00C2\u00BBU + 2.8 + 2.3 (-)1.5 (-)0.6 (+)0.9 + 3.3 (-)0.7 (-)0.6 (+)0.2 (-)0.7 E f f e c t o f WZ(-) on E - ^ N - 2.8 - 2.3 (+)1.5 , (+)0.6 (-)0.9 - 3.3 (+)0.7 (+)0.6 (- )0.2 (+)0.7 Note: F i r s t row estimates are seasonally uncorrected, second row estimates are seasonally c o r r e c t e d . -87-TABLE 6 Con't SIGNS OF COEFFICIENTS AND t VALUES AGE-SEX WOMEN GROUP 20-24 Years 25-34 Years 35-44 Years 4 5-54 Years 55 Years + Older Effect of UT(+) (+)0.4 (+)0.2 (-)0.3 (+)0.5 (-)0.4 on U\u00E2\u0080\u0094~Z (+)0.5 (+)0.4 (\")0.9 (-)O.O (-)0.5 Effect of UT(-) (-)0.6 (\")1.3 (-)0.7 (-)0.6 (-)0.8 on U\u00E2\u0080\u0094V ( - ) l . l - 2.1 (-)0.8 (-)0.5 <->1.6 Effect of WZ (+) (+)1.0 (+)1.5 (+)0.8 (-)0.1 (+)0.5 on E (+)0.5 + 2.4 (+)1.1 (+)0.2 (-)0.6 Effect of WZ(-) (- )0. 2 C-)1.8 (- )0.8 (- )0.5 (-)0.5 on U\u00E2\u0080\u0094\u00C2\u00BB-N (- )0.1 -2.9 (~)1.2 (~)0.8 (+)0.2 Effect o CD(-) (+)0.4 . (+)0.4 (+)1.3 (+)0.2 (+)1.1 on N\u00E2\u0080\u0094s\u00C2\u00BBN (+)1.1 (+)0.8 + 2.1 (+)0.1 + 2.0 Effect of u T ( - ) - 2.1 (+)1.2 (-)0.2 (+)0.7 - 2.5 on N\u00E2\u0080\u0094\u00C2\u00BB-U - 3.8 (+)0.5 (-)0.7 (-)0.6 - 3.2 Effect of u T ( + ) (+)0.9 (-)0.6 (-)O.O (~)0.3 (+)1.0 on N-\u00C2\u00BBN (-)O.l (-)0.9 (+)0.5 (-)O.l + 1.5 Effect of WZ(+) (+)1.8 + 3.7 + 3.7 + 3.9 + 2.1 on N-*E + 2.0 + 3.2 + 2.9 + 3.2 + 3.7 Effect of WZ(-) (~)0.7 - 4.4 - 3.3 - 3.6 <-)1.2 on N-\u00C2\u00BB-N (-)0.9 - 4.0 (-)1.5 - 3.0 - 2.5 Note: F i r s t row estimates are seasonally uncorrected, second row estimates are seasonally corrected. -88-TABLE 6 Con't AVERAGE ELASTICITIES AGE-SEX GROUP WOMEN 20-24 Years 25-34 Years 35-44 Years 45-54 Years 55 Years + Older % Change In E/P 0.1020 0.0760 0.2726 0.2113 0.2203 0.1465 0.2033 0.1378 0.0864 0.1350 Abso lute Change In E/P 0.0005510 0.0004106 0.0007738 0.0007548 0.0008357 0.0005559 0.0007663 0.0005202 0.0001385 0. r.00214 9 Z Change i n U/P 1.2576 1.2239 1.1302 1.6457 0.7032 2.1971 0.7634 1.0922 0.3806 0.0081 Abso lute Change i n U/P 0.0003463 0.0003770 0.0001359 0.0001978 0.0000735 0.000227 9 0.0000636 0.0001022 0.0000131 0.0000006 Z Change i n N/P -0.2076 -0.1729 -0.1759 -0.1510 -0.1490 -0.1284 -0.1360 -0.1015 -0.0185 -0.0258 Abso lute Change i n N/P -0.0008773 -0.000747 2 -0.0011095 -0.0007524 -0.0009092 -0.0007835 -0.0008352 -0.0006224 -0.0001570 -0.0002151 Z Change i n U/(E+U) 1.0978 1.0908 0.9272 1.3342 0.4687 1.9920 0.5459 0.9298 0.2354 -0.1214 Absolute Change In uV(E-HJ) 0.0005324 0.0025290 0.0002693 0.0004506 0.0001257 0.0005301 0.0001272 0.0002250 0.0000823 0.0000507 Note: F i r s t row estimates are seasonally uncorrected, second row estimates are seasonally corrected. The responses of women to incentives are not seasonal In nature. I base this conclusion on the insensitivity of the estimated coefficients for the economic variables to the inclusion of seasonal dummy variables. On the other hand, the men's response to changing incentives i s largely seasonal. This Interpretation i s based on the reduction i n the number of significant estimated coefficients for the economic variables when the model includes seasonal dummy variables. Without seasonal correction, 21 coefficients for women and 15 coefficients for men have the expected sign and are s t a t i s t i c a l l y significant at the 95% level. With seasonal correction, 24 coefficients for women and only 5 coefficients for men have the expected sign and are s t a t i s t i c a l l y significant at the 95% le v e l . These results suggest that women generally respond to changing labour force incentives while men only respond to changing incentives in a \"short-term\" seasonal pattern. The reductions of significant coefficients for men when seasonal dummy variables are included means that non-seasonal labour force behavior seems to be insensitive to the type of fluctuation i n incentives reflected in this study. These observations are consistent with the view that men are typically \"primary\" wage earners and women are mainly \"secondary\" wage earners in their families. Primary wage earners can not afford to adjust their \"long-term\" plans to minor fluctuations in labour force incentives. Secondary wage earners have satisfying alternatives to labor force participation and can withdraw without depriving their families of essential income. -90-Decisions by Workers Employed The e f f e c t of unemployment Insurance on the d e c i s i o n of workers to withdraw v o l u n t a r i l y from employment i s modeled as a fu n c t i o n of the \"replacement wage\". The \"replacement wage\" Is the r a t i o of the expected unemployment Insurance b e n e f i t to the expected wage l e v e l . T h i s r a t i o i s measured by the v a r i a b l e , V. I expect that when the unemployment insurance benefit increases r e l a t i v e to the return to employment the p r o b a b i l i t y of a worker q u i t t i n g w i l l increase. For workers who leave employment, e i t h e r through q u i t s or through l a y o f f s , the d e c i s i o n to remain i n the labour for c e i s modelled as a func t i o n of the expected return to labour for c e p a r t i c i p a t i o n fUs measured by the v a r i a b l e WZ, which i s the sum of the r e a l wage and the expected r e a l unemployment insurance b e n e f i t . I a n t i c i p a t e that the higher the value of WZ the greater the p r o b a b i l i t y that an unemployed worker would choose to remain In the labour for c e and search f o r employment. Table 6 shows that the estimated effects of V and WZ on the t r a n s i t i o n from employment i n period \" t \" to the three labour force s t a t e s i n period \"t+ 1 \" do not completely support the a n t i c i p a t e d pattern. The influence of V on men has the a n t i c i p a t e d e f f e c t when estimated without seasonal c o r r e c t i o n , but the c o e f f i c i e n t s with seasonal c o r r e c t i o n have the \"wrong\" sign i n the younger age groups f o r the t r a n s i t i o n from employment to \"not In the labour f o r c e \" . The -91-influence of V on women is consistent with the anticipated effect for the results with and without seasonal correction except for the transition from employment to unemployment in the middle age groups. The influence of WZ on men has the opposite effect with and without seasonal correction to that anticipated by the model. The results for the influence of WZ on women are mixed and inconclusive. In attempting to account for the actual results, i t Is best to examine V and WZ joi n t l y , since both are functions of the same variables: the unemployment insurance benefit and the wage. Any suggestion offered to explain these departures of the regression results from theoretical expectations are of course speculative. The variable V i s modelled as an influence on the decision to quit, but as indicated i n chapter 3, i t can also be Interpreted as Influencing layoffs by Implicit contracts as suggested by Feldstein (1976, 1978). There are two ways of viewing the results for WZ. Fi r s t the coefficients of WZ can be viewed as reflecting the decision whether to stay in the labour force for both the men who quit their jobs and those who are laid off. Alternatively the coefficients of WZ can be interpreted as measures of the responses of only those workers employed in period \" t \" who are involuntarily l a i d off. As mentioned ln chapter 3, i t i s conceivable that on average those who quit decide on withdrawal from the labour force differently from those who are la i d off. The results for V and WZ without seasonal correction taken together indicate that among men, job search (E\u00E2\u0080\u0094\u00C2\u00BB-U) i s increased when V rises -92-and WZ f a l l s . This effect i s particularly strong In the two youngest age groups (see tables 2.1, 2.2, and 2.3; and 3.1, 3.2, and 3.3). The simultaneous occurrence of these two movements i n the variables V and WZ means that the wage is f a l l i n g . Thus these findings indicate that a f a l l i n the wage stimulates job search and a rise in the wage discourages job search. The combined result suggests that among men when wages r i s e , dissatisfaction leading to job search is reduced, while higher earnings increase the demand for leisure. When one examines the results without seasonal correction for V and WZ rising and f a l l i n g together, the view that WZ influences only those l a i d off seems plausible. When V and WZ both rise, unemployment insurance benefits must be rising. The effects of V and WZ on (E\u00E2\u0080\u0094\u00C2\u00BBU) are now i n opposite directions: a rise in V raises (E\u00E2\u0080\u0094>-U) and lowers (E\u00E2\u0080\u0094>N) while a rise in WZ reduces (E\u00E2\u0080\u0094*-U) and raises (E\u00E2\u0080\u0094\u00C2\u00BB\u00E2\u0080\u00A2 N). These results are more plausible i f the two variables are interpreted as reflecting the behaviour of different groups. One may conjecture that the coefficients for V reflect the behaviour of those inclined to quit in order to search for a better job. They w i l l be more inclined to do so when higher unemployment insurance benefits reduce the \"penalty\". One may also conjecture that the coefficients for WZ reflect the behaviour of the marginal less \"motivated\" workers who are more l i k e l y to be l a i d off. They are more inclined to choose leisure when higher unemployment insurance benefits reduce the cost of leisure. The variable WZ was modelled to explain the effect of expected wage income and future employment insurance benefits, but the r e s u l t s suggest that WZ may be a measure of current income from these sources. Of course, the problem discussed i n Chapter 4 associated with i n t e r p r e t a t i n g the c o e f f i c i e n t s f o r WZ s t i l l remain. The above d i s c u s s i o n has been r e s t r i c t e d to the r e s u l t s f o r men without seasonal c o r r e c t i o n . For the regressions with seasonal c o r r e c t i o n the main d i f f e r e n c e from the above r e s u l t i s that the e f f e c t of V i s now ambiguous and not s i g n i f i c a n t . The i n t e r p r e t a t i o n of WZ i s the same. The r e s u l t s f o r women d i f f e r from those f o r men. The c o e f f i c i e n t s f o r WZ are ambiguous while the c o e f f i c i e n t s f o r V suggest a s i g n i f i c a n t pattern of behaviour. When I examine the r e s u l t s f o r the v a r i b l e V with and without seasonal c o r r e c t i o n , women appear to respond to higher unemployment insurance benefits by leaving the labour force (E\u00E2\u0080\u0094>-N). The influence of the v a r i a b l e V on the movement of women to unemployment (E >-U) i s unstable and i n s i g n i f i c a n t . In response to a r i s e i n V men appear to move i n t o unemployment (E\u00E2\u0080\u0094>-U) mainly on a seasonal basis (as shown by the greater strength of the r e s u l t s without seasonal c o r r e c t i o n ) . For women a r i s e i n V increases both seasonal and non-seasonal movements out of the labour force. These r e s u l t s suggest that there may be some substance to the charge that women treat unemployment as a \"retirement allowance\". Our r e s u l t s do not resolve the debate between the job search hypothesis proposed and the temporary l a y o f f hypothesis. Both approaches suggest that workers w i l l v o l u n t a r i l y leave employment f o r -94-suggest that the behaviour expected by Marston and Fe l d s t e i n occurs among men mainly In a seasonal manner. It i s quite l i k e l y that some workers would quit and i n t e n s i f y t h e i r search f o r better employment when unemployment insurance b e n e f i t s are Increased. . It i s equally p o s s i b l e that some workers and employers i n seasonal i n d u s t r i e s would e x p l o i t increased unemployment insurance b e n e f i t s through mutually acceptable temporary l a y o f f s ( i m p l i c i t c o n t r a c t ) . In summary the following p o s s i b i l i t i e s are suggested. Men l a i d o f f are more l i k e l y to choose l e i s u r e the higher t h e i r wage and/or unemployment Insurance income. Those i n c l i n e d to q u i t and job search are more l i k e l y to do so when unemployment insurance benefits are high and when wages are low. The r e s u l t that the young are more l i k e l y to fo l l o w these patterns i s not s u r p r i s i n g i n l i g h t of s e n i o r i t y preference s t r u c t u r e s i n the work place. Young men sampling a v a i l a b l e job opportunities may use unemployment insurance f o r job search a f t e r l e a v i n g a l e s s d e s i r a b l e seasonal job. The concentration of t h i s phenomenon among the young who do not have the career investment or family r e s p o n s i b i l i t i e s of the old e r \"prime age male\" workers Is a reasonable r e s u l t s . Decisions by Workers Unemployed The d e c i s i o n s confronting an unemployed worker are p a r t i a l l y determined by a v a i l a b l e job o f f e r s . The unemployed worker who has received a job o f f e r must choose to take the job (U\u00E2\u0080\u00949*E) or r e j e c t the job o f f e r . Those unemployed workers who do not receive a job o f f e r or turn down an unacceptable job o f f e r must decide whether to continue job search (U\u00E2\u0080\u0094*\u00C2\u00BBU) or leave the labour force (U\u00E2\u0080\u0094*-N). These decisions -95-have been modelled as functions of s e v e r a l v a r i a b l e s . In the model's s p e c i f i c a t i o n only some of the c o e f f i c i e n t s f o r the unemployment rate (U ) and the expected return to future employment (WZ) can be signed on the basis of t h e o r e t i c a l considerations. The d e c i s i o n of an unemployed worker not to accept an a v a i l a b l e job o f f e r and not to withdraw from the labour force (U\u00E2\u0080\u0094*-U) i s modelled as a function of the t o t a l unemployment rate. The t o t a l unemployment T rate i s measured by v a r i a b l e , U . I a n t i c i p a t e that an increase i n the rate of t o t a l unemployment increases the competition among the unemployed of each age-sex group f o r the a v a i l a b l e jobs. This i s expected to r e s u l t i n a greater w i l l i n g n e s s to accept scarce jobs. On the other hand, a r i s e i n the t o t a l unemployment rate i s expected to reduce the marginally unemployed worker's expectation of being successful i n h i s job search. These workers may become discouraged and leave the labour force u n t i l the job search climate improves f o r unemployed workers. The combined influences of the t o t a l unemployment rate on the d e c i s i o n of unemployed workers to remain unemployed between periods i s therefore expected to be negative. That i s to say when the t o t a l unemployment rate r i s e s , unemployed workers are expected to be more w i l l i n g to accept a v a i l a b l e jobs on the one hand, and f o r other unemployed workers who are unsuccessful i n t h e i r job search to leave the labour force through the \"discouraged worker\" eff e c t. The pattern f o r women f i t s the expected behaviour i n each case. Only one c o e f f i c i e n t i s s t a t i s t i c a l l y s i g n i f i c a n t at the 5% l e v e l . The pattern f o r men, however, i s mixed with only four out of ten - 9 6 -c o e f f i c i e n t s having the a n t i c i p a t e d sign. None of the c o e f f i c i e n t s f o r men are s i g n i f i c a n t at the 5% l e v e l . The d e c i s i o n of an unemployed worker to accept employment (U\u00E2\u0080\u0094*-E) T i s a l s o modelled as a function of U . I a n t i c i p a t e that an increase T i n U would increase the t r a n s i t i o n (U >-E) through a greater' w i l l i n g n e s s on the part of unemployed workers with job o f f e r s to accept. Consequently, the sign of the c o e f f i c i e n t s are expected to be p o s i t i v e . The pattern for men and women i s i r r e g u l a r . For both men and women h a l f of the c o e f f i c i e n t s are negative and h a l f are p o s i t i v e . A l l the c o e f f i c i e n t s are s t a t i s t i c a l l y i n s i g n i f i c a n t at the 5% l e v e l . The d e c i s i o n of an unemployed worker to leave the labour force i s T als o modelled as a fun c t i o n of U . As mentioned above I a n t i c i p a t e d T an increase i n U would increase the wi l l i n g n e s s to accept a job o f f e r , and at the same time i t would cause discouraged job searching workers to leave the labour force. The combined e f f e c t on (U\u00E2\u0080\u0094>*N) can not be determined a p r i o r i . The leakage to accepted job o f f e r s from T an increase i n U may o f f s e t the increased l i k e l i h o o d of 'discouraged workers\" le a v i n g the labour force. This leaves the expected sign of T the c o e f f i c i e n t on U f o r (U\u00E2\u0080\u0094>N) uncertain. T When the r e s u l t s of the e f f e c t s of U on ( U \u00E2\u0080\u0094 ^ E ) and (U\u00E2\u0080\u0094>-U) are combined I conclude that the behaviour of men i s not s i g n i f i c a n t l y a f f e c t e d . Women on the other hand appear to be influenced by T T U through a discouraged worker e f f e c t . The e f f e c t of U on the de c i s i o n of unemployed women to accept employment (U\u00E2\u0080\u0094>-E) i s not conclusive, yet the e f f e c t on (U\u00E2\u0080\u0094*-U) i s negative. I would conclude - 9 7 -then that the e f f e c t of U on unemployed women i s a discouraged worker effect.Examining the c o e f f i c i e n t s f o r (U-\u00C2\u00BB-N) i n Tables 2.6 and 3.6 1 f i n d t h i s conclusion i s weakly supported by seven out of ten c o e f f i c i e n t s f o r (U-\u00C2\u00BB-N) having a p o s i t i v e s i g n with only one p o s i t i v e c o e f f i c i e n t s t a t i s t i c a l l y s i g n i f i c a n t at the 5% l e v e l . The d e c i s i o n of an unemployed worker to accept employment ( U \u00E2\u0080\u0094 ^ E ) i s also modelled as a function of the v a r i a b l e , WZ, which measures the expected return to labour force p a r t i c i p a t i o n both d i r e c t l y through wages and i n d i r e c t l y through future unemployment insurance benefits. I a n t i c i p a t e that WZ w i l l influence the unemployed worker's d e c i s i o n * to accept job o f f e r s p o s i t i v e l y . The r e s u l t s vary between the groups examined. In general the expected behavior i s suggested for men under 3 5 years of age, f o r men 55 years and older, and f o r women under 45 years of age. The exclusion of prime age men (35 to 54 years old) i s consistent with the view that t h i s group does not make labour force decisions based on short-term f l u c t u a t i o n s i n incentives. The exclusion of \"older\" women (45 years and older) from the expected behavior may be e m p i r i c a l l y weak because very few women i n th i s age group are unemployed. The numerically important t r a n s i t i o n s f o r women i n these age groups appear to be between employment and \"not i n the labour force\". The d e c i s i o n of an unemployed worker to leave the labour force ( U ^ ^ N ) a f t e r not r e c e i v i n g a job o f f e r or r e j e c t i n g an unacceptable job o f f e r i s a l s o modelled as a function of WZ. I a n t i c i p a t e that - 9 8 -when the expected return to employment (WZ) f a l l s , marginal unemployed workers w i l l withdraw from the labour force i n favour of a c t i v i t i e s outside the labour market. The r e s u l t s i n Table 6 suggest that the expected behavior holds f o r women (with the exclusion of women over 54 years o l d when seasonal c o r r e c t i o n i s included). This i s consistent with e a r l i e r observations of the behavioral responses of women. E a r l i e r t h i s behavior was a t t r i b u t e d to the \"secondary worker\" phenomenon. The prime age men (35 to 54 years old) again appear unresponsive to minor f l u c t u a t i o n s i n labour f o r c e incen t i v e s . The young men (under 35 years old) respond to changes i n WZ i n a seasonal pattern only. This behavior i s a \" r a t i o n a l \" response to the a v a i l a b i l i t y of seasonal employment f o r younger workers who presumably have lower s e n i o r i t y and consequently a lower p r o b a b i l i t y of f i n d i n g a job during seasonal lows. For these workers an increase i n the expected return to employment should induce some to continue to search f o r work rather than leaving the. labour force u n t i l the regular seasonal upturn i n the job market. Older men (55 years and older) are generally responsive to changes i n WZ. I would i n t e r p r e t t h i s r e s u l t as the behavior of \" q u a s i - r e t i r e d \" men which must be s i m i l a r to that of \"secondary earners\". Because t h i s group contains a large proportion of the workers who decide to leave the labour force permanently and have b u i l t up resources f o r that purpose, the d e c i s i o n to extend t h e i r stay i n the labour f o r c e because of p o t e n t i a l gains should be s e n s i t i v e to f l u c t u a t i o n s i n incentiv e s . This group already has menbers with \"short-term\" commitments to labour -99-force p a r t i c i p a t i o n , therefore \"short-term\" responses to inc e n t i v e s should be expected. Decisions by Person Not i n the Labour Force The decisions confronting a person outside the labour force are p a r t i a l l y determined by the decisions of employers to make jobs a v a i l a b l e . The person must f i r s t decide to search for a job, and when he f i n d s an a v a i l a b l e job he must decide whether to take i t o r continue searching. These decisions have been modelled as functions of s e v e r a l v a r i a b l e s . Using the theory s e c t i o n , some of the signs f o r T the c o e f f i c i e n t s f o r the v a r i a b l e s , CD, U , and WZ can be a n t i c i p a t e d . The d e c i s i o n of a person outside the labour force to enter the labour force was modelled as a function of the change i n e f f e c t i v e demand, measured by the v a r i a b l e , CD. The influence of a r i s e i n CD on the t r a n s i t i o n from \"not i n the labour f o r c e \" to unemployment works through the decisions of workers to enter the labour force (tending to r a i s e unemployment), the decisions of workers r e c e i v i n g job o f f e r s to \"hold-out\" f o r b e t t e r o f f e r s (tending to r a i s e unemployment), and the d e c i s i o n of employers to increase job o f f e r s (tending to lower unemployment), therefore I can not assign a p r i o r i an expected e f f e c t of CD on ( N \u00E2\u0080\u0094 s * U ) . The influence of a r i s e i n CD on the t r a n s i t i o n from \"not i n the labour force\" to employment works through the same ambiguous process, therefore I can not assign a p r i o r i an expected e f f e c t of CD on ( N \u00E2\u0080\u0094 ^ E ) e i t h e r . The e f f e c t of CD on the t r a n s i t i o n from not i n the labour force i n period \" t \" to hot i n the labour force i n period \"t+1\" (N \u00E2\u0080\u0094*-N) can however be a n t i c i p a t e d . Here I have argued that an increase i n demand w i l l encourage workers -100-to enter the labour f o r c e , therefore reducing the t r a n s i t i o n (N-*-N). The uncertainty i s whether an increase i n CD w i l l Increase (N-*-U) or (N-&-E) or both. For men the expected e f f e c t on (N\u00E2\u0080\u0094>-N) Is observed f o r a l l age groups except the oldest on a seasonally corrected b a s i s , and f o r a l l age groups except the youngest and oldest on a seasonally uncorrected basis , though generally the e f f e c t i s not strong enough to be s t a t i s t i c a l l y s i g n i f i c a n t . Although the great majority of workers i n the prime age groups are l a r g e l y i n s e n s i t i v e to f l u c t u a t i o n s i n Incentives (except f o r seasonal quits or temporary l a y o f f s of the Fe l d s t e i n v a r i e t y ) , there appear to be marginal workers i n a l l age groups whose labour force p a r t i c i p a t i o n decisions are influenced by the a v a i l a b i l i t y of jobs. It i s also of i n t e r e s t that the estimated e f f e c t of CD i n th i s case i s one of the few instances i n which the seasonally corrected r e s u l t s f o r men give l a r g e r regression c o e f f i c i e n t s i n absolute magnitude than the r e s u l t s without seasonal dummies. Responses to c y c l i c a l rather than seasonal changes i n labour demand are therefore being observed. The estimated c o e f f i c i e n t s f o r the oldest men and the youngest men (seasonally uncorrected) are s t a t i s t i c a l l y i n s i g n i f i c a n t . They could, however, suggest a weak \"secondary worker\" e f f e c t , l i k e that of women, discussed below. The female groups a l l have the opposite r e a c t i o n to changes i n CD to that a n t i c i p a t e d . When e f f e c t i v e demand increases they stay out of -101-the labour f o r c e . This can be interpreted as a secondary worker phenomenon. When l a y o f f s are r i s i n g , women \"not i n the labour force\" e l e c t to search f o r work. This pattern Is consistent with the view that some women engaged i n non-labour for c e a c t i v i t i e s enter the labour force when other members of t h e i r family become unemployed. This r e a c t i o n Is an attempt to compensate f o r reduced family income. The d e c i s i o n of workers \"not i n the labour f o r c e \" to enter the labour force and accept job o f f e r s i s also modelled as a function of the t o t a l unemployment rate (U^). The e f f e c t of U^- on the d e c i s i o n to enter the labour force i s expected to be negative. The T increased competition i n d i c a t e d by a higher U i s expected to discourage entry. When increases, the t r a n s i t i o n (N\u00E2\u0080\u00943\u00C2\u00BB-N) should increase through reduced entry. The r e s u l t s f o r men and women are mixed, but a s t a t i s t i c a l l y s i g n i f i c a n t \"discouraged worker\" e f f e c t f o r the youngest and oldest men seasonally uncorrected and f o r the oldest men seasonally corrected occurs. The same pattern occurs f o r women but i s not s t a t i s t i c a l l y s i g n i f i c a n t at the 5% l e v e l . These r e s u l t s suggest that f o r some o l d and young workers t h e i r disadvantage i n f i n d i n g jobs increases as UT r i s e s . The t r a n s i t i o n from \"not i n the labour f o r c e \" to unemployment (N\u00E2\u0080\u0094-2>- U) i s al s o dependent on U^. When r i s e s , we a n t i c i p a t e that labour f o r c e entry w i l l decline, that those who do enter w i l l be more l i k e l y to accept job o f f e r s , and that employers w i l l be encouraged to o f f e r jobs to f i n d better workers among the unemployed. -102-Thls process should reduce the number of persons moving t o unemployment from \"not i n the labour f o r c e \" . Hie r e s u l t s f o r men are mixed and inconcl u s i v e f o r the e f f e c t of on ( N\u00E2\u0080\u0094*-U). The youngest and oldest women behave as an t i c i p a t e d with s t a t i s t i c a l l y s i g n i f i c a n t c o e f f i c i e n t s whether seasonally corrected or uncorrected. Women between 25 and 54 years o l d appear to have no s i g n i f i c a n t pattern. We i n t e r p r e t these r e s u l t s as evidence that members of the youngest and the oldest women groups outside the labour f o r c e respond to U*- as \"discouraged workers\". The t r a n s i t i o n ( N \u00E2\u0080\u0094 > \u00E2\u0080\u00A2 E) i s not expected to respond to U T i n a pr e d i c t a b l e way because U T i s expected to influence the decisions to enter the labour force and to accept a job o f f e r i n opposite ways. The d e c i s i o n of workers \"not i n the labour f o r c e \" to enter the labour f o r c e i s also modeled as a fun c t i o n of the expected return to employment, WZ. The e f f e c t of WZ on the worker's d e c i s i o n to enter the labour f o r c e i s expected to be p o s i t i v e , and therefore a negative e f f e c t on ( N \u00E2\u0080\u0094 i \u00C2\u00BB - N ) i s expected. The r e s u l t s f o r men are mixed and inc o n c l u s i v e , but the r e s u l t s f o r women have the expected s i g n and are s t a t i s t i c a l l y s i g n i f i c a n t at the 5% l e v e l i n s i x of the ten cases. Again the contrast between men and women i n response to f l u c t u a t i o n s In labour force incentives i s obtained. The combined e f f e c t of WZ on job acceptance and labour force entry i s expected to have a p o s i t i v e e f f e c t on the t r a n s i t i o n (N-\u00E2\u0080\u0094>\u00E2\u0080\u00A2 E ) . -103-Again the r e s u l t s f o r men are mixed while the r e s u l t s f o r women have the expected sign and are highly s i g n i f i c a n t . Nine out of ten c o e f f i c i e n t s f o r women are s t a t i s t i c a l l y s i g n i f i c a n t at the 5% l e v e l . The e f f e c t , of WZ on the the t r a n s i t i o n (N\u00E2\u0080\u0094*-U) i s not expected to have a p r e d i c t a b l e s i g n because i t s e f f e c t on labour force entry and job acceptance are opposing. Estimates of Average E l a s t i c i t i e s of Responses to Changes i n Unemployment Insurance I have discussed the estimated c o e f f i c i e n t s of the equation system. In t h i s section, I w i l l integrate these r e s u l t s and estimate the impact of a small change i n the unemployment insurance b e n e f i t s on the d i s t r i b u t i o n of each cohort of the population between the three labour force states (E/P, U/P, and N/P) and on the unemployment rate (U/ E + U ). The average e l a s t i c i t i e s and absolute changes f o r these four aggregate measures have been c a l c u l a t e d given a one percent increase i n the unemployment Insurance benefit l e v e l . These c a l c u l a t i o n s were made at the average value of the exogenous v a r i a b l e s and the lagged dependent v a r i a b l e s on the r i g h t side of the equations. Previous a n a l y s i s of the influence of unemployment Insurance on the labour market has focused on i n d i v i d u a l measures of the labour market (such as: unemployment, unemployment duration, turnover r a t e s , and labour force p a r t i c i p a t i o n ) . The s p e c i f i c a t i o n s of the determining v a r i a b l e s f o r these labour market measures should depend upon an integrated theory. Instead the equations f o r these labour market measures were s p e c i f i e d as s i n g l e equation models i n e a r l i e r work. -104-Consequently these studies have s p e c i f i c a t i o n s which are not s t r i c t l y consistent. For example, i f labour force p a r t i c i p a t i o n i s influenced by unemployment insurance, that influence should a l s o be s p e c i f i e d f o r employment, unemployment, and \"not i n the labour force\" through an integrated approach to d e c i s i o n making. This study's more integrated approach therefore provides a basis f o r re-evaluating e a r l i e r work. According to my model the response that each age-sex group makes to a small change i n the unemployment insurance benefit l e v e l depends upon the d e c i s i o n s by workers i n a l l three labour force states i n the previous period as well as the decisions of employers. When unemployment insurance b e n e f i t s increase, employed workers may e l e c t to quit t h e i r current employment or agree to temporary l a y o f f s . Unemployed workers may be more l i k e l y to accept employment because the expected return to employment has r i s e n , or else they may prolong or even discontinue job search when benefits r i s e . F i n a l l y , workers \"not i n the labour force\" may be induced by an increase i n the expected return to employment to enter the labour f o r c e . The net e f f e c t of these decisions on the three labour force states depends upon the previous d i s t r i b u t i o n of the population between labour force states and the impact of unemployment insurance on worker decisions. Table 6 includes the r e s u l t s of a 1% increase i n the unemployment insurance benefits at the average values of the exogenous variables and the lagged endogenous v a r i a b l e s f o r the f u l l time period of the a n a l y s i s . The proportion of the population employed increases for -105-women i n a l l age groups when the unemployment insurance benefits are increased, while the results d i f f e r for men i n di f f e r e n t age groups. The proportion of the population unemployed Increases f o r a l l women and declines f o r a l l men when the unemployment insurance benefits are increased. The proportion of the population \"not i n the labour force\" declines for a l l women and rises f o r a l l men when the unemployment insurance benefits are increased. The unemployment rate declines f o r men i n a l l age groups and ri s e s for a l l women except those over 54 years o l d , when seasonal correction i s used. Conclusions Grubel, Maki, and Sax (1975a) concluded that \"higher [unemployment insurance] benefits increased the welfare of the insured but cause losses i n society's aggregate output of market goods because of the induced unemployment\" This conclusion i s not supported by my re s u l t s , nor i s i t i n fact supported by t h e i r own r e s u l t s . Their work deals with the unemployment rate only and \"induced unemployment\" does not necessarily mean less employment. According to our model both employment, and unemployment, as well as the unemployment rate may r i s e i f workers are enticed into the labour force by the unemployment insurance change. Although we f i n d that d i f f e r e n t age-sex groups respond d i f f e r e n t l y to unemployment insurance changes, t h e i r response to an increase i n the unemployment Insurance benefits generally exerts upward pressure on the proportion of the population employed except f o r men 20 to 24 years old, and for men aged 25-34 and 55 years and older when seasonal correction i s used. These exceptions have a -106-r e l a t i v e l y small negative e f f e c t on employment compared to the estimated p o s i t i v e e f f e c t s . The Grubel, Maki, and Sax conclusion i s unwarranted because t h e i r model i s designed to examine the p o t e n t i a l impact of unemployment insurance on the unemployment rate only, neglecting the e f f e c t on the other labour force s t a t e s . The micro simulation by Rea (1977) of the unemployment insurance changes a l s o concluded that the unemployment rate would increase. Although our r e s u l t s also suggest that women's unemployment rates would r i s e , Rea's study i s l i m i t e d to persons i n the labour force paying unemployment insurance premiums between 1966 and 1970 and consequently can not take account of the increased labour force p a r t i c i p a t i o n . Like Grubel, Maki, and Sax, Rea finds increased unemployment among the population. We a l s o found that some employed workers would leave employment, but our c a l c u l a t i o n s suggest that among most age-sex groups there i s s u f f i c i e n t replacement by new labour force entrants. Green and Cousineau (1976) estimate the impact of the unemployment insurance changes on the l e v e l of unemployment and the labour force p a r t i c i p a t i o n r a t e . They estimate that the 1971 changes In the Unemployment Insurance programme increased both unemployment and the p a r t i c i p a t i o n r a t e i n 197 2 and 1973. I pointed out i n the review of the l i t e r a t u r e above that t h e i r findings also imply a decline i n employment i n 1972, and an increase i n employment i n 1973. Although I choose not to o f f e r an explanation f o r t h i s \" f l i p - f l o p \" -107-r e s u l t , I note that i t i s , l i k e ours, i n d i r e c t c o n f l i c t with Grubel, Maki, and Sax regarding the impact on employment and output. My r e s u l t s suggest that when unemployment insurance benefits r i s e , p a r t i c i p a t i o n i n the labour force would r i s e f o r women and decline for men, and that increases f o r women are t y p i c a l l y larger than the declines f o r men. These r e s u l t s also suggest a general increase i n employment with the aforementioned exceptions which are of r e l a t i v e l y small magnitude. The r e s u l t s also suggest that there would be fewer unemployed men and more unemployed women when unemployment insurance i s increased. Admittedly, our results are not aggregated, nonetheless the derived \" f l i p - f l o p \" behavior implied by the r e s u l t s of Green and Cousineau i s not r e a d i l y explained since the unemployment insurance scheme di d not change between 197 2 and 1973. Fred Lazar (1978) proposed to examine the e f f e c t s of unemployment insurance on job search. His fi n d i n g s support the p o s i t i o n that the average duration of unemployment increased with the increased unemployment insurance b e n e f i t s . He a t t r i b u t e s t h i s to extended search by workers. It would be necessary to examine the willingness of unemployed workers to accept employment to reach a conclusion on voluntary extended job search. Our r e s u l t s would suggest that i f unemployment duration has increased i t i s more l i k e l y to be due to increased labour force p a r t i c i p a t i o n rather than to workers refusing job o f f e r s . Lazar's f i n d i n g s do not r e s u l t from an ana l y s i s which con t r o l s f o r changes i n p a r t i c i p a t i o n rates, consequently we would -108-suggest that h i s conclusions are consistent with ours and do not ne c e s s a r i l y i n d i c a t e increased voluntary job search. The Sharir and Kuch (1977) study of the impact of the unemployment insurance changes i n 1955 and 1971 suggests s i g n i f i c a n t Increases In -labour force p a r t i c i p a t i o n rates f o r most age-sex groups. My study supports t h i s f i n d i n g f o r women with the exception of t h e i r f i n d i n g a decrease f o r women over 64 years o l d . This group was included i n our group \"55 years and old e r \" . Our findings f o r the group \"55 years and old e r \" i s r e l a t i v e l y weaker ( i . e . smaller average e l a s t i c i t y ) than f o r the other groups. This may be due to the pattern found by Sh a r i r and Kuch f o r the groups of women over 64 years o l d . Their f i n d i n g s f o r men are mixed, with both p o s i t i v e and negative e f f e c t s on p a r t i c i p a t i o n r a t e s . They make no estimate f o r men 20-24 years o l d or men over 64 years because of \"unacceptable\" estimated c o e f f i c i e n t s f o r t h e i r unemployment Insurance v a r i a b l e s . I a l s o found that the unemployment insurance v a r i a b l e s had r e l a t i v e l y weaker e f f e c t s on men than women. However the estimates f o r men a l l suggest that male p a r t i c i p a t i o n rates decline with increased unemployment insurance b e n e f i t s . The r e g i o n a l study of Swan, MacRae, and Steinberg (1976) examined the impact of unemployment insurance on labour supply i n three Maritime Provinces. They conclude: \" F i r s t , the greater generosity of the unemployment insurance system has not decreased true labour force -109-p a r t i c i p a t i o n i n the Maritimes. Second, p a r t i c i p a t i o n has increased i n New Brunswick and Prince Edward Island and more so than In Canada as a whole. T h i r d , p a r t i c i p a t i o n has not decreased i n Nova S c o t i a \" ^ . These findings are supportive of the general conclusions of Sharir and Kuch, Green and Cousineau, and my work, that increased unemployment insurance increases labour force p a r t i c i p a t i o n . In summary, although my study generally agrees with e a r l i e r work that increased unemployment insurance benefits may increase measured unemployment and labour force p a r t i c i p a t i o n , I t also f i n d s that unemployment insurance benefits would increase employment by i n c r e a s i n g the expected return to labour force p a r t i c i p a t i o n . On t h i s point Grubel, Maki, and Sax i n c o r r e c t l y conclude that t h e i r findings suggest employment would de c l i n e . Green and Cousineau's r e s u l t s suggest a d e c l i n e i n employment i n 1972 and an Increase i n employment i n 197 3. They do not acknowledge t h i s Implication of t h e i r estimates. F i n a l l y , Lazar's conclusion that increased average duration i s due to Increased job search Is not supported by my f i n d i n g s . Although I d i d not examine duration of unemployment d i r e c t l y , I did not f i n d strong changes i n job acceptance behavior. Therefore I would suggest that increased unemployment duration be a t t r i b u t e d to Increased p a r t i c i p a t i o n rates which would i n turn increase the job queue and reduce the l i k e l i h o o d of any p a r t i c u l a r i n d i v i d u a l r e c e i v i n g a job o f f e r . This crowding e f f e c t would increase the average duration of unemployment by i t s e l f . -110-There are two generalizations one can make from the Canadian studies of the labour market and unemployment insurance. F i r s t l y , after the 1971 changes to the Unemployment Insurance Act some workers i n the labour force worked less than they would have without these changes. Secondly, after the changes some people worked and searched for employment more than they would have. While i t is clear that voluntary recipients of unemployment insurance may be better off, i t i s equally clear that others may have suffered welfare losses due to the increased duration of unemployment resulting from the additional people searching for employment. Because I found no evidence to suggest aggregate output did decline, I conclude- that the 1971 changes in the Unemployment Insurance Act \"purchased\" additional job search by new labour force entrants. Presumably these new entrants have reduced their production of \"non-market goods and services\". Consequently the 1971 changes may have increased market employment and labour force participation at the expense of some non-market a c t i v i t i e s . On the other hand, the increased unemployment consequent on the changes In benefits may well have increased the production of \"non-market goods and services\". Any attempt to judge the welfare effect of the changes in the Unemployment Insurance Act by estimating aggregate output i n the indirect sense i s bound to end up in Idle speculation. It also, In my view, uses the wrong welfare criterion. Like other policy measures designed to \u00E2\u0080\u0094111-\"help the poor\", unemployment insurance should be judged by an appropriate c r i t e r i o n of welfare. FOOTNOTE 1 Grubel, Maki, and Sax (1975a), p.188. 9 Swan, MacRae, and Steinberg, p.24. -112-BIBLIOGRAPHY A l l e n , R.G. D. (1966) Mathematical Economics (New York: St. Martin's Press, 1966). Barten, A.P. (1969) \"Maximum L i k e l i h o o d Estimation of a Complete System of Demand Equations\", European Economic Review, 1-1, F a l l 1969, 7-73. Berndt, E.R. and Savin, N.E. (1974) \"Estimation and Hypothesis Testing i n Singular Equation Systems with Autogregressive Disturbances\", Econometrica, 43-5/6, September/November 1975, 937-957. Brunk, H.D. (1965) An Introduction to Mathematical S t a t i s t i c s (New York: B l a i s d e l l Publishing Company, 1965). Burdett, K and Mortensen, D.T. (1977) \"On Optimal Search Strategies and L a y o f f s \" , I n s t i t u t e f o r Research on Poverty Discussion Papers, #441-7 7, November 1977. Cain, G.G. and Watts, H.W. e t . a l . (1973) Income Maintenance and Labor Supply (Chicago: Rand McNally College Publishing Company, 1973). Chapin, G. (1971) \"Unemployment Insurance, Job Search, and the Demand fo r L e i s u r e \" , Western Economic Journal, 9-1, March 1971, 102-107. Conrod, C and Kunin, R (197 5), Comment: C r i t i q u e of Herbert G. Grubel, Dennis Maki, and Shelly Sax \"Real and Insurance Induced Unemployment i n Canada\" Canadian Journal of Economic. (Vancouver: mimeo, 1975). Cragg, J.G. (1973) Wage Changes and Labour Flows i n Canada (Ottawa: P r i c e s and Incomes Commission, 1973). Denton, F.T. , Feaver, C H . , Robb, A.L. (1976) The Short-Run Dynamics of the Canadian Labour Market (Ottawa: Economic Council of Canada 1976). Ehrenberg, R.G. and Oaxaca, R.L. (1976) \"Unemployment Insurance, Duration of Unemployment, and Subsequent Wage Gain\", American Economic Review, 66-5, December 1976, 754-766. -113-F e l d s t e i n , M. (1976) \"Temporary Layoffs i n the Theory of Unemployment\" Journal of P o l i t i c a l Economy, 84-5, October 1976, 937-957. (1978) \"The E f f e c t of Unemployment Insurance on Temporary Layoff Unemployment\" American Economic Review, 68-5 December 1978, 834-846. Foster, J . I . (1973) \"The Behaviour of Unemployment and U n f i l l e d Vacancies: Great B r i t a i n , 1958-1971 A Comment\" The Economic Journal, March 1973, 192-201. Green, C. and Cousineau, J. (197 6) Unemployment i n Canada: The Impact of Unemployment Insurance (Ottawa: Economic Council of Canada, 197 6). Green, C. (1977) \"Labour Market Performance from an Employment Perspective\" Canadian P u b l i c Policy/Analyse de P o l i t i q u e s , 3-3, Summer/Ete 1977, 315-323. Grubel, H.G., Maki, D. , and Sax, S. (1975a) \"Real and insurance induced unemployment i n Canada\" Canadian Journal of Economic/Revue canadienne d'economique, 8-2, May/Mai 1975, 174-191. (1975b) \"Real and insurance induced unemployment i n Canada. A reply \"Canadian Journal of Economics/Revue canadienne d'economique, 8-4 November/Novembre 1975, 603-605. Grubel, H.G. and Walker, M. et. a l . (1978) Unemployment Insurance Global Evidence of i t s E f f e c t s on Unemployment (Vancouver: The Fraser I n s t i t u t e , 1978). Guj a r a t i , D. (1972) \"The Behaviour of Unemployment and U n f i l l e d Vacancies: Great B r i t a i n , 1958-71\" The Economic Journal, March 1972, 195-204. -114-Holt, C.C. (1970) \"How Can the Phi l l i p s Curve be Moved to Reduce Both Inflation and Unemployment\" Microeconomic Foundations of Employment and Inflation Theory (New York: W.W. Norton and Company Inc., 1970), 224-256. Johnston, J. (197 2) Econometric Methods (New York: McGraw-Hill Book Company, 1972). Kaliski, S.F. (197 5) \"Real and insurance-induced unemployment In Canada\" Canadian Journal of Economics/Revue canadienne d'Economlque, 8-4, November/Novembre 1975, 600-603. ~ Lazar, F. (1978) \"The Impact of the 1971 unemployment insurance revisions on unemployment rates: another look \"Canadian Journal of Economics/Revue canadienne d'Economique, 11-3, August/Aout 1978, 559-570. Lee, T.C., Judge, G.G., and Takayama, T. (1965) \"On Estimating the Transition Probabilities of a Markov Process \"Journal of Farm Economics, 47-3, August 1965, 742-762. Marston, S.T. (1975) \"The Impact of Unemployment Insurance on Job Search \"Brookings Papers on Economic Activity (Washington: The Brookings Institute, 1975), 13-60. Mortensen, D.T. (1977) \"Unemployment Insurance and Job Search Decisions \"Industrial and Labor Relations Review, 30-4, July 1977, 505-520. Osberg, L. (197 9) \"Unemployment Insurance Amendments\" Canadian Public Policy/Analyse de Politiques, 5-2, Spring/Printemps 1979, 223-235. Phelps, E.S. e t . a l . (1970) Microeconomic Foundations of Employment and Inflation Theory (New York: W.W. Norton and Company Inc., 1970). Pollak, A.R. and Wales, T.J. (1969) \"Estimation of the Linear Expenditure System\" \"Econometrica, 37-4 October 1969, 611-628. -115-Powell, A. (1969) \"Altken Estimators as a Tool i n A l l o c a t i n g Predetermined Aggregates \"American S t a t i s t i c a l A s s o c i a t i o n Journal, September 1969, 913-922. Rea, S.A. (1977) \"Unemployment insurance and labour supply: a simulation of the 1971 Unemployment Insurance Act \"Canadian Journal of Economics/Revue canadienne d'Economique, 10-2, May/Mai 1977, 263-278. S h a r i r , S. and Kuch, P.J. (1977) \"More on the e f f e c t of unemployment insurance: the impact on p a r t i c i p a t i o n rates and l a y o f f s i n Canada 1953-1974\" Mimeo of paper presented at 1977 Canadian Economic Assoc i a t i o n meeting i n Fredericton. S t a t i s t i c s Canada (61-005) Indexes of Real Domestic Product by Industry (Ottawa). S t a t i s t i c s Canada (62-010) Consumer Pri c e s and P r i c e Indexes (Ottawa). S t a t i s t i c s Canada (71-001) The Labour Force (Ottawa). S t a t i s t i c s Canada (73-001) S t a t i s t i c a l Report on the Operation of the Unemployment Insurance Act (Ottawa). S t a t i s t i c s Canada (1973) Consumer Income Micro Data F i l e s (Ottawa). Swan, N., MacRae, P., and Steinberg, C. (1976) Income Maintenance Programs: The E f f e c t on Labour Supply and Aggregate Demand i n the Maritimes (Ottawa: Economic Council of Canada, 1976). Toikka, R. S. (1976) \"A Markovian Model of Labor Market Decisions by Workers\" American Economic Review, 66-5, December 1976, 821-834. Unemployment Insurance i n the 70's (Ottawa: Queen's P r i n t e r f o r Canada, 1970). Woodland, A.D. (1978) \"Stochastic S p e c i f i c a t i o n and the Estimation of Share Equations\", U n i v e r s i t y of B r i t i s h Columbia Discussion Paper No. 78-15, A p r i l 1978. "@en . "Thesis/Dissertation"@en . "10.14288/1.0094941"@en . "eng"@en . "Economics"@en . "Vancouver : University of British Columbia Library"@en . "University of British Columbia"@en . "For non-commercial purposes only, such as research, private study and education. Additional conditions apply, see Terms of Use https://open.library.ubc.ca/terms_of_use."@en . "Graduate"@en . "Unemployment insurance and the distribution of workers between labour force states"@en . "Text"@en . "http://hdl.handle.net/2429/22263"@en .