LABOR MARKET CRISIS AT MID-LIFE?
A Report of the Employment Prospects of People Aged 40–49
Wing Suen May Tam
Hong Kong Institute of Economics and Business Strategy The University of Hong Kong
October 2000
CONTENTS
List of Illustrations . . . iii
Executive Summary . . . v
1. Introduction . . . .1
1.1 Scope of the Study . . . 3
1.2 Outline of the Report . . . 4
2. Employment and Unemployment among the Middle-Aged . . . .6
2.1 Analysis of General Household Survey Records . . . 6
2.2 Age and the Incidence of Unemployment . . . 7
2.3 Labor Force Participation . . . 10
2.4 The Rising Demand for Educated Workers . . . 14
2.5 Declining Labor Participation among the Less Educated . . . 19
2.6 Multivariate Statistical Analysis . . . 21
2.7 Probit Model of Labor Participation . . . 25
2.8 Summary and Implications . . . 27
3. Analysis of Other Employment Outcomes . . . 30
3.1 The Structure of Labor Earnings . . . 30
3.2 Changes in the Earnings Structure . . . 33
3.3 Distribution of Employment by Industries . . . 34
3.4 Occupational Changes . . . 37
3.5 Long Term Unemployment . . . 40
3.6 Summary of Findings . . . 46
4. The Views of Employers . . . 48
4.1 Interviews with Employers’ Associations . . . 48
4.2 The Conduct of the Interviews . . . 49
4.3 Employers’ Evaluation of the Prospects for Middle-Aged Workers . . . 50
4.4 Skills Adequacy and Knowledge Needs . . . 55
4.5 Training for Middle-Aged Employees: Provision and Problems . . . 56
4.6 Preferences for Government Assistance . . . 58
4.7 Summary of Interview Results . . . 60
5. Training Provision: Employers’ Preferences and Concerns . . . 62
5.1 The Survey on Manpower Training and Job Skills Requirements . . . 62
5.2 The Incidence of Formal Training Provision . . . 65
5.3 Skills and Knowledge Adequacy of Less Educated Middle-Aged Workers . . . 69
5.4 Employers’ Preferences and Concerns about Government Subsidy . . . .71
5.5 Summary of Survey Results . . . 76
6. How the Government Can Help . . . 77
6.1 Diagnosing the Labor Market Problems of the Middle-Aged . . . 77
6.2 Some Basic Parameters in the Choice of Public Policies . . . 78
6.3 Helping Middle-Aged Unemployed Persons . . . 81
6.4 Preventive Measures . . . 84
6.5 Is Wage Subsidy a Good Idea? . . . 88
6.6 Summary of Policy Recommendations . . . 89
LIST OF ILLUSTRATIONS
Figures
2.1 Decomposing Data into Trend and Transitory Components . . . 7
2.2 Trend Unemployment for Different Age Groups, Men . . . 8
2.3 Transitory Unemployment for Different Age Groups, Men . . . 10
2.4 Trend Unemployment for Different Age Groups, Women . . . .11
2.5 Transitory Unemployment for Different Age Groups, Women . . . 11
2.6 Trend Labor Force Participation Rates, Men . . . 13
2.7 Trend Labor Force Participation Rates, Women . . . 13
2.8 Distribution of Educational Attainment among the Middle-Aged . . . 15
2.9 Age, Education, and Trend Unemployment (Men) . . . 17
2.10 Age, Education, and Trend Unemployment (Women) . . . 17
2.11 Age, Education, and Labor Participation (Men) . . . 20
2.12 Age, Education, and Labor Participation (Women) . . . .21
2.13 Estimated Age-Unemployment Profiles, 1999 . . . 23
2.14 Relationship between Age and Labor Participation, 1999 . . . 25
2.15 Age-Participation Profiles for Men with Primary Education . . . 27
3.1 Age-Earnings Profiles for Different Education Levels, 1999 . . . 31
3.2 Proportion Employed in Manufacturing Industries . . . 36
3.3 Proportion in Production-Related or Low-Skill Occupations . . . 38
3.4 Real Wage Indices by Broad Occupational Groups . . . 39
3.5 Percent Long Term Unemployed, Men . . . 43
3.6 Percent Long Term Unemployed, Women . . . 45
5.1 Incidence of Formal Training Provided to All Employees . . . 66
5.2 Formal Training which Include Middle-Aged, Less-Educated Workers . . . 67
Tables
2.1 Number of Persons by Educational Attainment, 1999 . . . 15
2.2 Number of Unemployed Persons among Various Groups . . . 16
2.3 Unemployment Probabilities Relative to Benchmark . . . 24
2.4 Participation Probabilities Relative to Benchmark . . . .26
3.1 Growth in Real Labor Earnings, 1986–1996 . . . 34
3.2 Average Monthly Employment Earnings by Occupation, 1996 . . . 40
3.3 Number of Long Term Unemployed Persons among Various Groups . . . 42
4.1 Interviews with Employers Associations . . . 51
5.1 Distribution of Respondent Establishments by Industry and by Size . . . 64
5.2 Reasons for Not Training the Middle-Aged Workers . . . 68
5.3 Employers’ Concerns Regarding Subsidy to Hire Replacement Workers . . . . 73
5.4 Concerns Regarding Direct Training Subsidy to Employers . . . 74
EXECUTIVE SUMMARY
Background
• Conventional wisdom holds that the prime of one’s working life is near the age of forty. However, the pace of technological change at the turn of the new century seems to be faster than what we have experienced in the past several decades. As- sociated with this wave of new technology is a rapid transformation of the structure of the economy. Accelerated obsolescence of labor market skills and the use of new and knowledge-intensive methods of production may arguably have changed the re- lationship between age and productivity. These developments have led to mounting concerns about a decline in the employment prospects of middle-aged people.
• In 1999 there were 911,000 middle-aged (i.e., aged between 40 and 49) people who were economically active. They made up one quarter of the Hong Kong labor force.
Over half of them have low educational attainment (i.e., an education at the lower secondary level or below). They were over-represented in manufacturing industries and in craft, production, or elementary occupations. These industrial and occupa- tional sectors have been at jeopardy as Hong Kong shifts to a knowledge-intensive and service economy. Most of these workers have a family to support. The social implications of any decline in their employment prospects cannot be overlooked.
• Extant evidence on the employment problems of middle-aged people is mainly de- rived from anecdotes, which are likely to be imprecise and piecemeal. A systematic assessment of these problems is needed. This report makes use of comprehensive and territory-wide data to examine the labor market prospects of the middle-aged.
Particular attention is paid to middle-aged people who are less educated, that is, those with an education at the lower secondary level (Form 3) or below. This study draws on the following types of data:
records of the General Household Survey for every quarter in the period 1985–
1999;
the 1986, 1991, and 1996 population census and by-census records;
a special survey of a representative sample of employers conducted in May–June, 2000; and
in-depth interviews with a number of employer organizations.
Assessment of the employment prospects of the middle-aged
1. Middle-aged people generally fare no worse than people in other age groups in terms of the main indicators of employment prospects.
• We use a time-series technique in business cycle research to decompose the quarterly unemployment rate into a trend (i.e., long term) component and a transitory (i.e., temporary) component. Between 1985 and 1999, middle-aged men and women had a trend unemployment rate which was 0.7 or 0.8 percentage point below that of the general population. Moreover there is no evidence that this advantage has eroded in recent years despite structural changes in the economy.
• Transitory unemployment among the middle-aged is less volatile compared to that among the general population, which indicates that middle-aged workers on average are less vulnerable to business cycles than are other workers. People who are par- ticularly hard hit by the rise in unemployment in Hong Kong since 1990 are those in the 15–24 age group.
• The labor force participation rate of all middle-aged men has remained relatively stable at an average of 98 percent in the period 1985–1999. For middle-aged women, the average rate was at 53 percent, which was five percentage points higher than that of all women aged 15 or above.
• Middle-aged workers tend to have higher earnings than do workers in other age groups. Middle-aged workers with low educational attainment fare no worse than do other workers with low education in terms of earnings or earnings growth.
2. Among middle-aged people, those who are less-educated face deteriorating labor market prospects. This is due to their low education level rather than to their age.
• Since 1990, less-educated men and women in all age groups were increasingly more likely to be unemployed, compared to the overall male and female population. In tan- dem with this development, the trend increase in unemployment among middle-aged people with low educational attainment has been faster than that for the general population. The relative deterioration in the employment prospects of this group of individuals is attributable to their low education rather than to their age. Ris- ing demand for educated workers in Hong Kong has produced a bifurcation in the employment outcomes between skilled labor and less-skilled labor.
• Between 1985 and 1999, the trend labor force participation rate of less-educated middle-aged women decreased from 52 percent to 46 percent, while the overall fe- male labor participation rate remained stable at 49 percent. However, the drop in labor force participation among less-educated middle-aged women was not more pro- nounced than the drop in labor force participation among all less-educated women.
• In the same period, the trend labor force participation rate of less-educated middle- aged men decreased from 97 percent to 95 percent. Again, low educational attain- ment rather than age is the main culprit behind such decrease. The fall in labor force participation rate for all less-educated men is higher than that for less-educated men in the 40–49 age group.
3. Our study does not give middle-aged workers a completely clean bill of health.
There are two potential problem areas that are of concern.
• Middle-aged workers (especially men) with low educational attainment are more likely to be in production-related and low-skill occupations than are other workers.
Moreover they seem to have difficulty moving away from these declining occupations.
Because of their specific human capital, these workers are not the first to suffer as the demand for low-skill jobs falls. However, they do not face a bright prospect if the trend of stagnating wages in low-skill jobs does not improve.
• Although the absolute number of middle-aged people who are unemployed for six months or more is still small, these individuals make up an increasing fraction of unemployed middle-aged persons. Moreover their difficulties in finding new jobs are partly due to the age factor.
4. Less-educated middle-aged employees are unlikely to receive formal training from their employers. These employees are deemed to lack specific work skills and knowledge.
• A majority of Hong Kong companies do not provide their employees with formal job training. The incidence rate is particularly low in small establishments, which lack the capability to offer formal job training. Even when training is provided, less- educated middle-aged employees are not particularly likely to be recipients. Em- ployers seem to rely primarily on informal job training instead of formal training.
• Employers believe that the relocation of manufacturing production to mainland China and China’s entry into the World Trade Organization will increase the need for the management and marketing skills of middle-aged employees who are in line or middle management. For middle-aged employees in low level jobs, the perceived financial returns from investing in formal training are low. Employers also doubt the motivation of middle-aged employees to attend regular formal training sessions.
They generally prefer training younger workers to training middle-aged workers.
• Employers do recognize that less-educated middle-aged employees lack specific work skills and knowledge and would like to see improvements if they do not have to bear the full cost. Computer skills are generally preferred by employers. Other skills such as occupational safety, sales and customer relations, and interpersonal communication skills at work are also in demand.
• In terms of the preference for government assistance in training provision, the overall orientation of employers is minimal involvement on their side and least disruption to the normal functioning of their business. Pecuniary training costs are of secondary importance.
How the government can help
1. The scale of the problem calls for prudent and focused policy measures.
• Our assessment of the employment prospects of middle-aged people indicates that the scale of the problem is smaller than a labor market crisis as suggested by anecdotal evidence. Prudent and focused policy responses are, therefore, more appropriate than across-the-board drastic remedial actions.
• Remedial policies which substantially raise the cost of doing business will only do more harm than good. We do not favor providing currently employed workers with direct training subsidies or offering wage subsidies to new recruits because they are not cost effective.
• Our assessment identifies two problem areas: (1) middle-aged workers who have difficulty in switching away from low-skill occupations; and (2) the long-term unem- ployed among the middle-aged. The existing employees retraining program, which targets at helping these people, should continue to be supported and strengthened.
Unemployed middle-aged persons have low opportunity costs of time and have a high risk of suffering from prolonged unemployment. Focusing government resources on helping this group of individuals is more effective than a diffused effort to help all middle-aged persons.
2. A few positive and focused measures can be adopted to alleviate pockets of potential problem areas for the middle-aged.
• Vocational learning certification system for the unemployed
The existing employees retraining program can be strengthened by introducing a system of vocational learning certification. This will provide employers with credible and relevant information about labor quality. It will also raise the incentive of the middle-aged people to acquire training. The certificates should be competence based rather than attendance based.
Active participation by employers in the design of the certification system is
of paramount importance. Inputs should be obtained from the Education and Manpower Bureau, the Employees Retraining Board, the Vocational Training Council, established training bodies, and human resources professionals. Close monitoring of training bodies and standardized training materials are needed to ensure uniformity and quality.
In the initial stage, the certification system can center on four main types of skills preferred by employers: (1) use of computer systems and software; (2) occupational safety and health; (3) sales and customer relations skills; and (4) interpersonal communication skills at work.
• Annual training leave for the currently employed
The government can encourage or require employers to provide employees with two to three days of annual training leave. This will relieve the time constraints of employees and lower the cost of training borne by them. These training leaves should not be based on age lest employers would substitute away from middle- aged workers.
The training can be conducted in a modular approach. Training leaves can only be used toward attaining qualifications under the proposed vocational learning certification system. Each module should last for two to three days so as to encourage attendance.
• Publicity campaign to raise awareness for self-improvement
The government needs to promote the proposed training leave and certifica- tion system through a publicity campaign. This will help to make workers aware of labor market developments, the need to learn new skills and knowl- edge, the virtues of saving for adversity, and the importance of continuous self- improvement. When these policy measures are implemented as a set, they will gradually overcome the reluctance of middle-aged workers to acquire training for their own betterment.
3. Expanding education opportunities is the long-term solution.
• Tackling the employment problems of middle-aged workers is a task that requires prudence and patience. Our diagnosis indicates that many of the labor market problems of the middle-aged stem from their inadequate education. Expanding education opportunities and improving the quality of education are the generic and long-term solution to these problems.
㆗年㆟的就業出現危機？
㆗年㆟的就業出現危機？
㆗年㆟的就業出現危機？
㆗年㆟的就業出現危機？
40-49 40-49 40-49
40-49 歲㆟士就業前景報告 歲㆟士就業前景報告 歲㆟士就業前景報告 歲㆟士就業前景報告
孫永泉、譚若梅
香港大學香港經濟與商業策略研究所 2000 年 10 月
摘要 摘要 摘要 摘要
背景 背景背景 背景
z 以往，40 歲左右是㆒個㆟㆒生㆗的事業黃金時期。但新世紀來臨後，科技發展得比過去 幾十年都要快，並帶動經濟結構急速轉型，導致個㆟的勞動技能迅速過時，加㆖新的生產 方式改用知識密集，這㆒切都改變了年齡和生產力的關係，令㆟擔心㆗年㆟士的就業前景 將不如前。
z 香港去年有 911,000 名從事經濟活動的(economically active)「㆗年㆟士」（指年齡介乎 40-49 歲者），佔整體勞動力㆕分之㆒。但其㆗逾半屬「低教育程度」（指㆗㆔或以㆘程度 者），而且偏重於製造業，工種方面偏重於工藝、生產和非技術職業。但隨著香港的經濟 轉為知識密集型和服務業為主，這些行業和職業日趨式微。㆗年㆟士大都需要養家，他們 就業前景惡化的社會後果不容忽視。
z 目前，有關㆗年㆟士就業困難的例證大都基於傳聞，支離破碎，有欠準確。故此，有需要 對有關問題作有系統的評估。本報告分析全面的全港數據，試圖有系統及全面㆞了解㆗年
㆟士的勞動市場前景，尤其是低教育程度㆟士的困難。所採用的數據包括：
1985 年至 1999 年各季度的《綜合住戶統計調查》；
1986 年、1991 年和 1996 年的㆟口統計和㆗期㆟口統計(by-census)；
2000 年 5 月至 6 月間㆒項對全港僱主進行的抽樣調查；以及
對㆒批僱主組織的深入訪問。
對㆗年㆟士就業前景的評估 對㆗年㆟士就業前景的評估對㆗年㆟士就業前景的評估 對㆗年㆟士就業前景的評估
1. 就各種主要的就業前景指標來說，㆗年㆟士的情況㆒般並不比其他年齡的㆟士差。
本報告以商業週期研究㆗常用的「時序數列」(time series)方法，分析各季度的失業率，分 解箇㆗的長期趨勢（trend，亦即長線關係）和過渡性變化（亦即臨時關係）。從 1985 年至 1999 年，㆗年㆟士的長期趨勢失業率較整體㆟口平均值低 0.7-0.8 個百分點。雖然近年經濟 出現結構性的變化，但並無證據顯示，㆗年㆟士的這項優勢有所削弱。
㆗年㆟士的過渡性失業率亦不若整體㆟口起落大，這顯示與其他年齡的㆟士相比，㆗年勞動
㆟口平均來說受商業週期的影響較小。事實㆖，自 1990 年起，每當失業率㆖升時，影響最大 的是以 15-24 歲的㆟口。
1985 年至 1999 年間，㆗年男性的勞動參與率(labor force participation rate)相對穩定，
平均值維持在 98%。㆗年女性同期內的平均值為 53%，較全港年滿 15 歲女性的平均值高 5 個 百份點。
㆗年勞動者的收入也高於其他年齡組合。無論就收入水平或收入增長率來說，低教育程度㆗
年勞動者的情況並不比其他年齡的低教育程度勞動者為差。
2. 就㆗年㆟士來說，低教育程度者的就業前景有惡化的趨勢。但原因是他們的教育程度低，
與年齡無關。
1990 年以來，不管是那個年齡，低教育程度㆟士的失業可能性都有所㆖升。與此並行的是，
低教育程度㆗年㆟士的長期失業趨勢較整體㆟口㆖升得快。㆗年㆟口的就業前景惡化，原因 在於教育程度低而不於年齡。由於市場需要較高教育水平的工㆟，技能水平高和沒有技能的 工㆟就有 截然不同的就業情況。
1985 年至 1999 年間，女性的勞動力參與率穩定於 49%的水平。低教育程度㆗年女性的長期趨 勢勞動力參與率由 52%㆘降到 46%；但㆗年女性的參與率跌幅並不比其他年齡的女性為大。
同期，低教育程度的㆗年男性的長期趨勢勞動參與率由 97%㆘降到 95%。與女性的情況相同，
㆘跌的主要是因為教育程度低而非年齡。事實㆖，所有低教育程度男性的勞動參與率跌幅反 而高於 40 到 49 歲低教育程度㆟士。
3.但本研究發現，㆗年勞動者的就業前景並非㆒片光明。有兩個問題是值得關注：
與其他年齡的勞動者相比，低教育程度的㆗年㆟仕（尤其是男性）較多從事生產性或低技能 職業。而且，這類低教育程度的㆗年男性似乎較難離開衰退的行業。由於低教育程度的㆗年 男性具有某種特定的㆟力資本，即使在低技能職位的需求減少時，但他們未必即時失業。但 若低技能職位的工資長期呆滯，低教育水平的㆗年男性的就業前景堪虞。
失業半年或以㆖的㆗年㆟士為數不多，但他們佔整體失業㆗年㆟士㆗的比例日漸㆖升。而且，
他們轉業時遇到困難的原因與他們的年齡相關。
乏某些工作技能和知識。
香港大多數企業沒有為僱員提供正規的職業訓練。其㆗以小企業為甚，因為他們根本沒有能 力這樣做。而僱主即使提供培訓，也不㆒定提供給受教育程度較低的㆗年僱員。僱主似乎主 要依靠非正規而不是正規的在職訓練。
香港僱主認為，隨著製造業北移到內㆞、以及㆗國加入「世界貿易組織」(WTO)，將會需要更 多有管理和開發市場技能的僱員，對於㆗層管理的㆗年僱員需求會有所增加。至於為處於低 級職位的㆗年僱員提供正規培訓，㆒方面投資於他們身㆖，可見的回報偏低。另㆒方面，僱 主亦懷疑這類㆗年僱員參予正規培訓課程的動力，故此，㆒般選擇培訓較年輕的僱員。
僱主認為，教育程度較低的㆗年僱員缺乏某些技能和知識，故只要不須負擔培訓的全部費用，
僱主會希望僱員能改善他們的技能及知識，特別是電腦技術、職業安全、銷售、與顧客的關 係和在工作場合與別㆟溝通的技能。
對於由政府協助提供培訓，僱主總的來說有以㆘的考慮：盡量少麻煩僱主﹔盡量不影響公司 的正常運作。是否由政府來分擔培訓的費用反屬次要。
政府可以如何幫忙㆗年㆟仕 政府可以如何幫忙㆗年㆟仕政府可以如何幫忙㆗年㆟仕 政府可以如何幫忙㆗年㆟仕
1. 由於問題的規模，政策措施須慎重周密、目標明確。
本報告的評估顯示，㆗年㆟士的就業前景尚未如傳聞般陷入「勞動力市場危機」。故此，在製 定政策時，應採取慎重周密、目標明確的方法，而非㆒刀切(across-the board)的大動作。
有關政策以不會大幅度增加經營成本為宜，否則將弊多於利。由於直接津貼接受培訓的現有 僱員或對新聘僱員提供工資津貼並不符合成本效益，故此本報告並不推薦。
本報告發現有兩類㆟仕的就業問題較嚴重：(1)從事低技能職業的㆗年勞動者轉行有困難﹔及 (2)㆒些長期失業的㆗年㆟士。政府現有的職業再培訓計劃以這兩類㆟仕為對象，故此該計劃 有需要進㆒步加強，以協助這兩類㆟仕。這些㆗年失業者有較低的時間機會成本，他們亦有 可能長期失業。因此，若能集㆗資源協助這兩類㆟仕，收效會比分散資源，協助所有㆗年㆟
士為佳。
2. 可同時採取以㆘積極和目標明確的措施，舒緩㆗年㆟仕其他遭遇到的就業問題。
為失業㆟士設立職業進修證書制度
z 設立職業進修證書制度以加強現有的再培訓計劃
這制度可為僱主對有關求職者的質素，提供可信和適用的資訊，同時，亦會誘發㆗年㆟仕接 受培訓。但頒授證書應按照接受培訓者的能力，而不應取決於出席率。
制訂證書制度時，務必鼓勵僱主積極參與，並徵求教育及㆟力統籌局、僱員再培訓局、職業 訓練局、往績良好的培訓組織、以及㆟力資源專家的意見。同時密切監察各培訓組織和標準 的培訓教材，確保水準劃㆒、質素良好。
證書制度初期宜集㆗提供以㆘㆕種僱主認為僱員欠缺的技能：(1)電腦系統和軟件的應用；(2)
每年為現有僱員提供培訓薪假
政府應鼓勵或規定僱主，每年為僱員提供兩、㆔㆝的培訓有薪假期，以免僱員因為㆖班而不 能接受培訓，同時降低培訓對僱員構成的成本。培訓薪假不應與年齡掛鉤，以免僱主藉機辭 退㆗年僱員。
有薪培訓可採用模塊方式推行，但只適用於㆖文建議的職業進修證書制度內的幾種資格。每 個模塊為期兩、㆔㆝，以鼓勵僱員選修。
發動宣傳，提高市民對自我完善的認識
政府宜發動宣傳，推廣㆖文建議的培訓薪假和證書制度，讓市民了解勞動市場的進展、掌握 新技能和知識的需求、以及不斷自我改善的重要性。當以㆖措施能㆒整套㆞實施，可望逐漸 消除㆗年勞動㆟口對接受培訓才能自我完善的猶疑。
3.增加教育機會方為長遠解決之道
解決㆗年勞動力的就業問題需要慎密的考慮和耐心。本報告的分析顯示，不少㆗年㆟士就業 有困難，主要的原因是他們的教育程度不足，故此擴大接受教育的機會，改善教育質素是治 根及長遠之計。
（完）
LABOR MARKET CRISIS AT MID-LIFE?
A Report of the Employment Prospects of People Aged 40–49
Chapter 1 Introduction
Conventional wisdom has it that the prime of one’s work life is near the age of forty.
At that point in the life cycle, physical conditions and the state of the intellect are still well suited for productive work, while the maturity and the skills accumulated from prior work experience bear their fruits in the labor market. This conventional view is supported by casual observation as well as by a massive amount of data. Economists and other social scientists have produced literally thousands of studies that investigate the relationship between age and market wage or earnings. These studies cover many different countries—large and small, rich and poor. They span many different time periods—from the nineteenth century to the end of the twentieth century. The unifor- mity of of their conclusion is striking. Almost invariably, they find that labor market earnings rise steadily with age for the first twenty years after a person leaves school, and begin to decline gradually after a person reaches age fifty. The age-earnings profile reaches a plateau around age forty. Is there reason to believe that this pattern will change at the dawn of the new century? Will there be a labor market crisis looming over workers in their forties, who are supposed to be at the peak of their productive life?
Concern over the employment prospects of workers in their forties is not completely unwarranted. The pace of technological change at the turn of the new century seems to be faster than what we have experienced in the past several decades. Associated with this wave of new technology is a rapid transformation of the structure of economy.
It is now fashionable to argue that we are entering a “new economy,” in which the conventional wisdom of economists is no longer sacred. Some of the claims made about the new economy are just hype, but some others cannot be simply dismissed.
Accelerated obsolescence of labor market skills and the use of new and knowledge- intensive methods of production may arguably have changed the relationship between
age and productivity. Ten years ago, a chief executive officer younger than fifty was a rarity. The average chief executive officer of a major U.S. corporation was 56 or 58 years old.1 Today, in some quarters of the internet business, it is difficult to find anyone older than forty in the executive board room. At the middle management level and below, the concept of lifetime employment has given way to corporate down-sizing and restructuring. Long tenure in a firm is no longer a guarantee for job security. In the United States, there is evidence that workers’ subjective beliefs about their job security have turned more pessimistic,2 although the evidence is mixed on whether job stability has actually deteriorated or not.3
At the lower end of the skills spectrum, structural changes in the economy can potentially have devastating effects on middle-aged workers. These workers may have accumulated some human capital specific to their current employment. But when the industries for whom they work are in decline, or when the type of jobs they do are replaced by different ones, the human capital they possess will no longer be so valuable.
Moreover, economist Theodore Schultz points out that formal education helps make a person more adaptable and more capable of dealing with temporary dislocation.4 Since a significant fraction of these low skill workers have little schooling, they may not be well equipped to face the possibility of a labor market crisis in mid-life.
Any deterioration in the employment prospects of people in their forties can have a major impact on the Hong Kong economy. In 1999, there were 911,000 people aged between 40 and 49 who were economically active. They make up over a quarter of
1 Sunita Wadekar Bhargava, “Portrait of a CEO: What’s the Typical Boss Like? Here are the Vital Statistics,” Business Week, October 11, 1993, p. 64; Michael J. McCarthy, “A CEO’s Life: Money, Security and Meetings,” Wall Street Journal, July 7, 1987, p. 27.
2 Stefanie R. Schmidt, “Long-Run Trends in Workers’ Beliefs about Their Own Job Security: Evi- dence from the General Social Survey,” Journal of Labor Economics, 17, October 1999, Part 2, pp. S127–
S141.
3 Henry S. Farber, “The Changing Face of Job Loss in the United States, 1981–1995,” Brookings Papers on Economic Activity: Microeconomics, 1997, pp. 55–128; David Neumark, Daniel Polsky, and Daniel Hansen, “Has Job Stability Declined Yet? New Evidence for the 1990s,” Journal of Labor Economics, 17, October 1999, Part 2, pp. S29–S64; Robert G. Valletta, “Declining Job Security,”
Journal of Labor Economics, 17, October 1999, Part 2, pp. S170–S197.
4 Theodore W. Schultz, “The Value of the Ability to Deal with Disequilibria,” Journal of Economic Literature, 13, September 1975, pp. 827–846.
the Hong Kong labor force. Of these 911,000 individuals, over half—or 477,000—have an education at the lower secondary level or below. These 40–49 year-old workers with relatively low levels of education are over-represented in manufacturing industries, which have shown a long term trend of decline in Hong Kong. They are also more likely than the average worker to be working in crafts, production, or elementary occupations, which have come under stress as the economy shifts toward more knowledge-intensive modes of operation. The mere size of this group of middle-aged low-skilled workers suffices to call the attention of policy makers. Furthermore, most of these workers have a family to support. The social implications of any decline in their employment prospects cannot be overlooked.
1.1 Scope of the Study
This report examines the labor market prospects of people aged between 40 and 49. For convenience they are referred to as “middle-aged workers,” even though not all of them are currently employed. Particular attention will be focused on middle-aged workers with low levels of education, i.e., those at the lower secondary level (Form 3) or below.
The employment prospects of middle-aged workers is a subject of much speculation.
Anecdotes abound but systematic evidence is not so common. This study makes use of records from the General Household Survey for every quarter in the period 1985–
1999. The General Household Survey is a regular survey conducted by the Census and Statistics Department based on a random sample of households covering over 70,000 individuals in Hong Kong. It provides a comprehensive database for evaluating the labor market problems faced by middle-aged workers. Occasionally this database is supplemented by the 1986, 1991, and 1996 population census and by-census records. In contrast to other surveys and studies of individual industries or occupations, these two sources of information are much larger in scale and are representative of the territory as a whole. And because the General Household Survey follows a consistent protocol over time, labor market trends can be discerned more easily using information from this survey than using other sporadic sources of information.
Whereas the General Household Survey and the population censuses contain exten-
sive information about the characteristics of employees, they reveal almost nothing about the firms. To obtain a more balanced view from both sides of the labor market, a special survey of firms was conducted in April–July 2000 for the purpose of this study. This sur- vey was carried out with logistical support from the Census and Statistics Department.
It covers a representative sample of all employers in Hong Kong. In addition to factual information about their employment practices, the survey also solicits employers’ views on their needs and preferences regarding training provisions to middle-aged workers with low levels of education. This structured survey is supplemented by in-depth interview with a number of employer organizations. The views collected form an important input in the formulation and evaluation of public policy options.
1.2 Outline of the Report
We begin this study by taking a quantitative examination of the labor market situation of middle-aged workers based on records from the General Household Surveys for the period 1985–1999. Chapter 2 looks at two important labor market outcomes: labor force participation and the incidence of unemployment. We use time series techniques to isolate the temporary fluctuations from the more permanent trends so as to arrive at a prognosis of the employment prospects of middle-aged workers. We pay particular attention to separating the effects of age and educational attainment on employment and unemployment.
Chapter 3 continues to examine other labor market outcomes for the middle-aged.
Labor earnings is a more permanent determinant of the standard of living than is un- employment. Therefore we focus on the structure of labor earnings and its changes over time. The second half of this chapter tries to identify particular problem areas by specifically looking at unemployed middle-aged workers and middle-aged workers who are in manufacturing industries and low-skilled occupations.
After the quantitative study of labor market trends, we turn to a qualitative study of these trends in Chapter 4. Employers are intimately knowledgeable about business conditions, which shape their demand for labor. They also provide training to employees, which affect their employment prospects. Chapter 4 reports the results of our intensive
interviews with representatives of employers’ associations. They provide useful insights into the skills adequacy of middle-aged workers and the problems of training provision.
The qualitative interviews with employers are supplemented by a more structured questionnaire survey conducted with the help of the Census and Statistics Department.
Details of the survey results are discussed in Chapter 5.
The final chapter of this report draws on the results from the earlier chapters to inform the choice of public policies. Specific policy recommendations are made with the objective of helping middle-aged workers. The limitations and constraints on effective intervention are also pointed out.
Chapter 2
Employment and Unemployment among the Middle-Aged
2.1 Analysis of General Household Survey Records
The General Household Survey provides a consistent and comprehensive database for assessing the employment situation in Hong Kong. Official statistics such as the un- employment rate and the size of the labor force are compiled from the results of this survey, which is conducted at quarterly intervals on a random sample of some 20,000 households. This report uses data from the General Household Survey for the period 1985–1999.
For each quarter in the period 1985–1999, relevant employment statistics are ex- tracted from the General Household Survey files. The time series thus constructed are rather volatile. Sampling error, temporary fluctuations, seasonal effects, and cyclical factors introduce considerable short term variability to the data series. In order to dis- tinguish the longer term movements from the short term variability, each time series is treated by a statistical method known as the Hodrick-Prescott filter.1
The Hodrick-Prescott filter is the most commonly used technique in business cycle research to separate trend movements from transitory fluctuations. Its use is illustrated in Figure 2.1, which plots the time series data when the Hodrick-Prescott filter is applied to the unemployment rate. The dotted line is the trend component of unemployment, and the dashed line is the transitory component. These two components add up to the original time series, shown in solid line of the figure.
Focusing on the trend component of the time series has two advantages for the purpose of this study. First, the trend component is smoother than the original series.
This makes the visual identification of data patterns much easier. Second, because
1 Robert Hodrick and Edward Prescott, “Postwar U.S. Business Cycles: An Empirical Investiga- tion,” Journal of Money, Credit, and Banking, 29, February 1997, pp. 1–16.
of policy lags, labor market policies should be made to address structural problems.
Transitory fluctuations in the labor market due to business cycles or other short term factors should not influence the formulation of public policies toward the structure of the labor market. In what follows, most of the time series data used in this report are the trend components of the original series extracted by using the Hodrick-Prescott filter, although some attention will also be given to transitory labor market movements.
2.2 Age and the Incidence of Unemployment
Unemployment in Hong Kong stood at an all time high of 6.2 percent in 1999. The burden of unemployment is not distributed evenly across the population. Teenagers and the unskilled typically have higher unemployment rates than other groups. While middle-aged workers tend to have relatively low unemployment rates, the costs of unem- ployment can be particularly high for this group. These workers often have worked in the same company for a long period. Being suddenly thrown out of work can be a traumatic experience, both economically and psychologically. The difficulties are compounded if
alternative employment opportunities are few and far between.
To investigate the relationship between age and the incidence of unemployment, we extract the trend components of the overall unemployment rate and of the unemployment rate for the 40–49 age group. For comparison, we also extract the trend unemployment rate for the 15–24 year-olds. Since men and women tend to exhibit different labor market behavior, they are analyzed separately. Figure 2.2 shows the pattern for men.
A close examination of Figure 2.2 leads to the following conclusions:
• Unemployment among 40–49 year-old men is consistently lower than overall male unemployment by approximately 0.7 percentage point. In contrast, men in the 15–
24 age group experience an unemployment rate which is on average 3.6 percentage points higher than overall male unemployment.
• The three unemployment series all show an upward trend since 1990. We have also looked at the unemployment rates for men in other age groups, and they exhibit
the same pattern. This suggests that the rise in unemployment since 1990 is an economy-wide problem rather than an age-related problem.
• The gap between the dotted line (aged 15–24) and the solid line (aged 15+) has widened since 1990. In 1990Q1, trend unemployment among men aged 15–24 was 2.3 percentage points higher than overall male unemployment. By 1999Q4, the difference has risen to 8.2 percentage points. On the other hand, the gap between the dashed line (aged 40–49) and the solid line (aged 15+) has remained more or less constant throughout the period. The group that are particularly hard hit by the rise in unemployment since 1990 are young people in the 15–24 age group, not middle-aged people in the 40–49 age group.
In Figure 2.3, we plot the transitory components of unemployment rate for men in different age groups. The figure shows that transitory unemployment among 40–49 year- old men (dashed line) follows the overall pattern (solid line) very closely. In contrast, transitory unemployment among 15–24 year-olds (dotted line) is much more volatile than the overall pattern. During a transitory upswing of the economy, business can expand rapidly by absorbing young workers. But when the economy is in a down-swing, these young workers are also the first to go. Employer-employee attachments tend to be stronger for middle-aged workers than for young workers. For most long-serving employees, both employers and the employees themselves typically have invested in building up firm-specific human capital. If employers lay off these workers in a temporary downturn, they will lose the opportunity to reap the benefits from their human capital investments when business picks up again. This explains why we observe in Figure 2.3 that the 40–49 age group enjoy greater employment stability than the 15–24 age group.
In Figures 2.4 and 2.5, we plot the trend and transitory unemployment rates for different groups of women. The patterns shown in these two figures are almost identical to the patterns we have observed for men. Unemployment rate among middle-aged women is lower and more stable compared to younger women. Figure 2.4 shows that middle-aged women has a trend unemployment rate which is 0.8 percentage point below the overall female trend unemployment rate in 1990Q1. Although this advantage has
narrowed somewhat to 0.6 percentage point in 1999Q4, the magnitude of the change is not significant enough to cause serious concern.
The preliminary evidence presented in Figures 2.2 to 2.5 suggests that conventional wisdom is well and alive: middle-aged people are less prone to unemployment risks than are people in other age groups. They have a lower unemployment rate than the population average. Moreover, notwithstanding the structural changes taking place in the Hong Kong economy, this advantage has not significantly diminished over time. If there is any reason to focus on the age factor as a possible problem in the labor market, it is the 15–24 year-olds who deserve more attention.
2.3 Labor Force Participation
The lack of gainful employment does not necessarily show up in unemployment statistics, because people who do not actively look for work are not counted as part of the labor force. To arrive at a more complete picture of the employment situation, it is useful
to study labor force participation rates as well as unemployment rates. Although the decision to join the labor force is often regarded as a supply response on the part of the workers, we believe that this decision is affected by demand conditions too.2 After all, being out of the labor force and being unemployed are not really so distinct empirically.3 This section is devoted to examining labor participation trends for people of different age groups.
Figure 2.6 plots the trend labor force participation rates for all men aged 15 or above (solid line) and for men in the 40–49 age group (dashed line). It can be seen that male labor force participation is trending steadily downward from 81 percent in the beginning of the sample period to about 75 percent toward the end of 1999. The primary reasons for this decline are the increase in schooling among the young and earlier retirement among the old. Among 40–49 year-old men, labor force participation has remained relatively stable at about 98 percent: almost all men in this age group are economically active.
A more careful look at Figure 2.6 reveals that the labor force participation rate of 40–49 year-old men has dropped slightly relative to that of 25–39 year-old men (shown in dotted line) since 1992. However, the magnitude of this drop is very small. The difference between these two series never exceeds 2 percentage points. There is no evidence of a substantial decline in labor participation among middle-aged men.
Women exhibit very different labor participation trends compared to men. In Figure 2.7, female labor force participation rate (solid line) averages to 48 percent during the 1985–1999 period. Labor force participation rate for women in the 40–49 age group (dashed line) is higher, at 53 percent, during the same period. While these two series are trending up in recent years, there is no noticeable upward or downward trend over the entire sample period. Moreover, the gap between these two series has not widened or narrowed.
2 Chuihui Juhn, “Decline of Male Labor Market Participation: The Role of Declining Market Opportunities,” Quarterly Journal of Economics, 107, February 1992, pp. 79–121.
3 James M. Poterba and Lawrence H. Summers, “Reporting Errors and Labor Market Dynamics,”
Econometrica, 54, November 1986, pp. 1319–1338
The most striking feature of Figure 2.7, however, is the secular increase in female labor participation among the 25–39 age group (dotted line). The labor force participa- tion rate among this group has risen by 13 percentage points over the past 15 years, from 59 percent in 1985 to 72 percent in 1999. Why didn’t 40–49 year-old women experience the same rapid increase in labor participation? In the next two sections, we will argue that the reason has more to do with education than with age.
2.4 The Rising Demand for Educated Workers
The evidence presented so far suggests that it is a mistake to believe that middle- aged workers are disadvantaged in the labor market: they have a lower than average unemployment rate and a higher than average labor force participation rate. By virtue of their age and therefore their years of accumulated experience in the labor market, these workers are relatively well-endowed with human capital. On-the-job training and more subtle learning-by-doing give them an advantage over younger and less experienced workers.
Human capital accumulated through training and learning-by-doing, however, is not the same as human capital acquired through formal education. The current 40–49 year-olds were born between 1951 and 1960, a time when education opportunities were less plentiful than they are now. For example, these individuals were past their school age when free and compulsory nine-year education was introduced in Hong Kong in 1978. In general, middle-aged people in Hong Kong are less well educated compared to younger generations or compared to the overall population. This is illustrated in Figure 2.8, which shows the percentage distribution of educational attainment for the 40–49 age group (light bars). The distribution of educational attainment for the general population aged 15 or above is also shown (dark bars) for comparison. In 1999, 48 percent of the Hong Kong population aged 15 or above had no more than nine years of education (i.e., lower secondary education or below). Among people aged between 40 and 49, on the other hand, as many as 57 percent had no more than nine years of schooling. More detailed figures are shown in Table 2.1.
There is growing evidence that technological changes in the past two decades have
Table 2.1
Number of Persons by Educational Attainment, 1999
Population Labor Force
aged 15+ aged 40–49 aged 15+ aged 40–49
no schooling/kindergarten 471,850 34,990 59,950 17,860
primary (P1–P3) 1,270,720 388,680 605,480 261,600
lower secondary (S1–S3) 933,400 256,520 634,570 197,570 upper secondary (S4–47) 1,942,090 357,140 1,398,350 292,740
tertiary (non-degree) 383,630 61,590 307,880 56,560
tertiary (degree) 594,450 91,980 470,330 84,940
been “skill-biased” in the sense that they raised the productivity of skilled workers relative to that of unskilled workers. This is manifested in increasing returns to educa- tion, stagnant wage growth for unskilled workers, and a trend of rising wage disparity.4
4 Lawrence F. Katz and Kevin M. Murphy, “Changes in Relative Wages, 1963–1987: Supply and Demand Factor,” Quarterly Journal of Economics, 107, February 1992, pp. 35–78; John Bound and George Johnson, “Changes in the Structure of Wages in the 1980’s: An Evaluation of Alternative
Table 2.2
Number of Unemployed Persons among Various Groups
aged 40–49 all individuals aged 40–49 & education≤ F.3
1985 83,620 8,760 6,450
(3.2%) (2.3%) (2.5%)
1990 36,550 3,850 2,430
(1.3%) (0.8%) (0.8%)
1995 95,630 18,080 14,430
(3.2%) (2.7%) (3.8%)
1999 217,060 49,090 34,140
(6.2%) (5.4%) (7.2%)
Lawrence Katz and Kevin Murphy have documented that people with less than high school education in the United States have experienced a 6.6 percent fall in real wages in the period 1979–1987. In contrast, people with college education or above have ex- perienced a 7.7 percent wage gain during that period. Chinhui Juhn has shown that the decline in male labor force participation in the United States from 1967 to 1987 was much greater among high school dropouts than among college graduates. Formal education has become increasingly important in today’s labor market.
The large concentration of the middle-aged in Hong Kong with little formal education is a cause for concern. While we have argued that the 40-49 year-olds are generally not a disadvantaged group in the labor market by virtue of their age, a significant fraction of these individuals may indeed be disadvantaged by virtue of their low educational attainment. We therefore need to break down the analysis by education as well as by age. Table 2.2 shows the number of unemployed persons and the unemployment rate (in parentheses) for various groups. As before, we find it more illuminating to perform an analysis of trends by using the Hodrick-Prescott filtering technique. Such an analysis is depicted in Figures 2.9 and 2.10, which show the trend unemployment rates for various demographic groups.
Explanations,” American Economic Review, 82, June 1992, pp. 371–392.
Figure 2.9 displays the unemployment trends for men and Figure 2.10 displays the unemployment trends for women. The visual patterns shown in these two figures for the 1990–1999 period are almost identical, so we can save some space by focusing on Figure 2.9 only. This figure reveals the following patterns:
• Men with an educational attainment of Form 3 or below (shown in dashed line) tend to have a higher unemployment rate than do the overall male population (thick solid line). The difference between these two groups’ unemployment rates was negligible, at about 0.1 percentage point, before 1990. The gap had widened substantially since then. By 1999, unemployment among the less well educated was 2.1 percentage points higher than that among the general population.
• A similar deterioration in unemployment for the less well educated can be detected within the age group 40–49. Before 1990, middle-aged men with an educational attainment of Form 3 or below (thin solid line) were only marginally more likely to be unemployed than all middle-aged men (dotted line). By the end of the sample period, however, the difference between their unemployment rates had grown to 1.9 percentage points.
• Middle-aged men with low educational attainment (thin solid line), however, have lower unemployment when compared to all men with low educational attainment (dashed line). The gap between the trend unemployment rates for these two groups is fairly stable at about 0.6 percentage point. This suggests that the rise in unem- ployment for middle-aged men with low educational attainment is not attributable to the age factor.
• Between 1985 and 1990, being less well educated had only a negligible effect on raising unemployment while being middle-aged would lower unemployment risks by 0.6 percentage point. Thus, middle-aged men with low educational attainment had a lower unemployment rate than the general male population. The thin solid line lies below the thick solid line during this period. Since 1990, being less well educated becomes increasingly a handicap in the labor market, while the advantage from being middle-aged remains unchanged. As a result, middle-aged men with low educational
attainment face a more rapid rise in unemployment rate when compared to the male population as a whole. The thin solid line starts from a position below the thick solid line and eventually rises above it. By the end of 1999, middle-aged men with low educational attainment had an unemployment rate which was 0.9 percentage point higher than the overall male unemployment rate.
In sum, our analysis suggests that unemployment risks are indeed deteriorating for middle-aged men and women with low levels of education. This deterioration in their labor market prospects, however, is largely attributable to their education rather than to their age. For the past ten years people with low educational attainment, regardless of their age, have been facing a worsening unemployment trend relative to the general population.
2.5 Declining Labor Participation among the Less Educated
In addition to the relationship between education and unemployment trends, we have performed a similar analysis to investigate the relationship between education and labor participation trends. Figure 2.11 plots the trend components of the labor force partic- ipation rates for various groups of men. It can be seen that men with low educational attainment (shown in dashed line) have a greater drop in labor force participation rate compared to all men (thick solid line). A large part of this is due to the fact that the former group is over-represented by elderly people. Since earlier retirement among el- derly people is a major factor behind the drop in male labor force participation rate in Hong Kong, the divergence between the dashed line and the solid line can be readily understood.5
Once age composition is held fixed, the effect of education on labor participation is much less pronounced. Among men in the 40–49 age group, for example, those with low educational attainment (thin solid line in Figure 2.11) have a labor force partici- pation rate which is only marginally below that of all middle-aged men (dotted line).
Thus, middle-aged men with low educational attainment still have an average labor force
5 Wing Suen, “Retirement Patterns in Hong Kong: A Censored Regression Analysis,” Journal of Population Economics, 10, October 1997, pp. 443–461.
participation rate of 97 percent in this period.
The labor force participation rate for this group of middle-aged men with low ed- ucational attainment did fall slightly in recent years, to 95 percent in 1999. Their employment has dropped relative to all middle-aged men (thin solid line versus dotted line). Relative to all men with low educational attainment, however, their employment has risen (thin solid line versus dashed line). One may conclude that any decline in the employment prospects of this group is more due to their education than to their age.
The labor force participation trends described here are even more evident for women than for men. Figure 2.12 shows the smoothed labor force participation rates for different groups of women. If we compare the thin solid line in this figure with the dotted line, it can be observed that middle-aged women with low educational attainment had significantly reduced their labor participation relative to all middle-aged women. The difference in their trend labor force participation rates grew from 1.3 percentage points to 8.5 percentage points. If we compare the thin solid line with the dashed line, on the
other hand, it can be observed that the same group of women did not reduce their labor participation relative to all women with low educational attainment.
However, the labor force participation rate for women with low educational attain- ment (dashed line in Figure 2.12) has declined rapidly relative to the general population (thick solid line), with the gap widening from 7 percentage points in the beginning of the sample period to 20 percentage points toward the end. Thus, even though middle-aged women with low educational attainment did not have a significant drop in labor force participation relative to all women with low educational attainment, they did have a significant drop in labor force participation relative to the overall female population.
Again, such a drop is attributable to the education factor than to the age factor.
2.6 Multivariate Statistical Analysis
To disentangle the effects of age and education on employment outcomes, we perform multivariate statistical analysis using the individual records from the General Household
Survey. Such an analysis will give a more precise estimate of the effects of various determinants of employment outcomes than is afforded by a visual inspection of data trends.
Since unemployment is a binary (“unemployed” or “not unemployed”) outcome, it can be modeled by a probit regression. We assume that the probability that a person is unemployed depends on his or her education (six categories), marital status (three categories), the quarter of the survey (four categories), and age. To build flexibility in modeling the effects of age, we enter age as a polynomial of degree four. Furthermore, this polynomial is interacted with a dummy variable indicating low educational attainment (Form 3 or below) so as to capture the different effects of age on the well educated and on the less well educated. The probit regressions are performed separately for men and women aged between 15 and 59 for the years 1985, 1990, 1995, and 1999.
The results of the probit regression estimates could be presented numerically, but they would be difficult to interpret. To better understand these results, we plot the predicted unemployment probability against age based on the regression models. Figure 2.13 shows two such plots for men and women for the year 1999. In both panels, the solid line represents people with primary education and the dotted line represents people with upper secondary education or matriculation. For men, unemployment risk declines rapidly with age until around age 30, after which unemployment risk rises slowly with age. Men with lower educational attainment are more likely to be unemployed than men with higher educational attainment. For women, the patterns are very similar, except that the rise in unemployment risk after age 30 is slower than that for men. The difference in unemployment risks between well educated women and less well educated women is also smaller than that for men.
Take a 35 year old married man with upper secondary education who was surveyed in the fourth quarter of 1999 as the benchmark for comparison. According to the probit model, his probability of being unemployed was 2.9 percent. Holding education constant while increasing the age to 45 raises the unemployment probability by 1.1 percentage points. Holding age constant while lowering the education to primary level raises the
unemployment probability by 4.4 percentage points. Finally, increasing the age to 45 while lowering the education to primary level raises the unemployment probability by 5.2 percentage points. These calculations are displayed in Table 2.3 for the years 1985, 1990, 1995, and 1999.
Rows (A2) and (B2) of Table 2.3 illustrate the effect of age on unemployment. One can see that raising the age from 35 to 45 raises the probability of unemployment, but the size of this effect is relatively modest—typically less than one percentage point. The effect of education on unemployment is shown in rows (A3) and (B3). A person with primary education is more likely to be unemployed than is a person with upper secondary education. Moreover, the gap between the well educated and the less well educated has grown considerably, reaching 4.4 percentage points for men and 1.9 percentage points for women in 1999.
Row (A4) of Table 2.3 shows the increase in unemployment probability for men when age is increased from 35 to 45 and education is lowered from upper secondary level to
Table 2.3
Unemployment Probabilities Relative to Benchmark
1985 1990 1995 1999
(A) Men
(A1) aged 35 and S4–S7 education (benchmark) 1.1% 0.3% 1.1% 2.9%
(A2) aged 45 and S4–S7 education LESS (A1) +0.6% +0.3% +0.4% +1.1%
(A3) aged 35 and P1–P6 education LESS (A1) +0.9% +0.3% +2.1% +4.4%
(A4) aged 45 and P1–P6 education LESS (A1) +1.9% +0.6% +2.7% +5.2%
(B) Women
(B1) aged 35 and S4–S7 education (benchmark) 0.6% 0.1% 0.9% 2.1%
(B2) aged 45 and S4–S7 education LESS (B1) +0.3% +0.1% +0.3% +0.6%
(B3) aged 35 and P1–P6 education LESS (B1) +0.8% +0.4% +2.1% +1.9%
(B4) aged 45 and P1–P6 education LESS (B1) +0.6% +0.2% +2.2% +2.7%
primary level. If the effect of being middle-aged and the effect of being less-educated do not compound together to produce a large adverse effect on unemployment, row (A4) should be approximately the same as the sum of rows (A2) and (A3). Table 2.3 shows that this is indeed the case. For the year 1995, the sum of the age effect and the education effect (2.5 percent) was less than the predicted excess unemployment for middle-aged men with primary education (2.7 percent) relative to the benchmark. For the year 1999, the reverse in true: the sum of the age effect and the education effect (5.5 percent) was greater than the predicted excess unemployment for middle-aged men with primary education (5.2 percent) relative to the benchmark. In both years, the difference between these predicted values are fairly small. A similar point can be made for women.
With the exception of year 1999, row (B4) is less than the sum of rows (B2) and (B3).
Our observation implies that the adverse effect on unemployment of being middle-aged is not more pronounced among people with low educational attainment than among the well educated. It also means that the adverse effect on unemployment of having low educational attainment is not more pronounced among the middle-aged than among 35 year-olds.
2.7 Probit Model of Labor Participation
We have also estimated probit models for the labor force participation decision. The independent variables used are the same as those used in the probit models for unem- ployment. Figure 2.14 gives a graphical representation of the estimated relationship between age and labor force participation for the year 1999.
Panel (A) of Figure 2.14 indicates that, beyond age 30, male labor force participation declines gradually with age among people with primary education. No such decline is observed among men with upper secondary education. For women, panel (B) of the figure shows that labor participation declines with age among the highly educated, while there is a double-peaked pattern among women with lower educational attainment. The depression between the two peaks can be explained by women’s withdrawal from the labor force due to child-bearing and child-rearing.
Table 2.4 provides some illustrative comparisons of labor force participation rates.
For men, labor force participation among people aged 35 and with upper secondary
Table 2.4
Participation Probabilities Relative to Benchmark
1985 1990 1995 1999
(A) Men
(A1) aged 35 and S4–S7 education (benchmark) 99.8% 99.9% 99.9% 99.8%
(A2) aged 45 and S4–S7 education LESS (A1) −0.6% −0.4% −0.5% −1.1%
(A3) aged 35 and P1–P6 education LESS (A1) −0.8% −0.5% −0.7% −2.3%
(A4) aged 45 and P1–P6 education LESS (A1) −2.5% −2.2% −2.4% −7.0%
(B) Women
(B1) aged 35 and S4–S7 education (benchmark) 57.6% 62.7% 68.9% 71.5%
(B2) aged 45 and S4–S7 education LESS (B1) −3.9% −3.8% −11.1% −16.0%
(B3) aged 35 and P1–P6 education LESS (B1) −15.4% −24.0% −33.0% −32.5%
(B4) aged 45 and P1–P6 education LESS (B1) −14.1% −19.2% −30.0% −25.9%
education stayed constant at close to 100 percent throughout the period. Compared to this benchmark group, people who are ten years older have a slightly lower labor participation rate, as shown in row (A2) of the table. Similarly, people who have primary education instead of upper secondary education also have lower labor participation rates, as shown in row (A3).
The most noteworthy feature of Table 2.4 is that the effect of age and education on male labor force participation has increased in recent years. For example, row (A4) shows that before 1995 middle-aged men with low educational attainment had a labor force participation rate which was only about 2 percentage points lower than the benchmark group. In 1999, however, the gap widened substantially to 7 percentage points.
Figure 2.15 plots the estimated relationship between age and labor participation for men with low educational attainment. The age-participation profiles are almost identical for the years 1985, 1990, and 1995. Moreover the slope of the profiles are almost flat between age 25 and age 50. In the year 1999, however, the age-participation profile has shifted down somewhat, and there is a perceptible negative slope between age 25 and age 50. This figure suggests that there is a deterioration in the employment prospects
of middle-aged men with low educational attainment relative to men who are younger and more educated.
For women, the most striking trend that may be observed from Table 2.4 is the rapid rise in labor participation among those aged 35 with upper secondary education (the benchmark group). In contrast, the labor participation rate among those aged 45 with primary education has fluctuated within narrow bands. For the latter group of women, there is no absolute decline in their employment prospects, but there is a relative deterioration nonetheless. At a time when labor participation is rapidly rising among educated women, middle-aged women with low educational attainment have failed to catch up.
2.8 Summary and Implications
The main findings presented in this chapter can be briefly summarized:
• Middle-aged men and women have a lower unemployment rate and a higher labor
