Download the entire document in Adobe Acrobat's Portable Document Format (PDF)
This report is based on the work carried out during 1997-98. Nomaan Majid of the Employment Sector of the ILO, Geneva is the coordinator and principal author of the report. The work is based on contributions from a team comprising of Arshad Zaman Associates, A.R.Kemal, M. Irfan, M. Mahmood, G. Chaudhry and Nomaan Majid. Particular thanks are due to A.K. Ghose, M. Mahmood and Akmal Hussain for detailed comments on earlier drafts. Thanks are also due to Samir Radwan, Rolph Van der Hoeven and Rashid Amjad for suggestions on the final draft. A.S.Oberai of ILO-SAAT, provided constant support and encouragement for the project from its inception. ILO-SAAT, the ILO Office in Islamabad and the UNDP in Islamabad actively facilitated the progress of this report, in its most critical phases. Very special thanks are due to them.
This report is an outcome of work done by the ILO, on an employment strategy for Pakistan during 1997 and 1998, an activity which was funded under the UNDP's SPPD facilty. One objective of the core work done for this exercise was to contribute an input to Planning Commission for their employment and manpower related work on the Ninth five Year Plan. Another objective was to produce a employment strategy document, that was cognizant both of sectoral employment issues and the macro economic environment in which a future employment strategy needs to be conceived. It is hoped that work is seen as a step in the right direction. More thinking however, needs to be done in detailing the proposed employment strategy, in addition to addressing the implications of this strategy for social groups within sectors. It is expected that the ILO in collaboration with the UN agency system will carry this work further.
Two central issues need to be emphasized. This document argues that in the Pakistan context the key to an employment growth revival is the revival of the manufacturing sector. The particular forms that such a revival may take, the linkages with other sectors, and some of the policy elements required for such a revival are delineated. The document also suggests in a separate section why the role of the macro policy environment in the growth process is critical from an employment perspective. It argues where and how Pakistan came to its fiscal crisis. This event, which is a result of imperatives on the part of State as well as ill-conceived reform sequencing on the part of donors, preceded the post nuclear situation of additional economic constraints that the country has faced and the more recent changes in government.
Today when the elements of adjustment packages and their relationship to growth are being seriously questioned by Bretton Woods institutions themselves because of the high social costs of adjustment as evidenced by the Asian and Russian crises, it is important to directly examine the nature of policy packages necessary for a revival of employment-friendly growth in developing countries. The challenge is not easy for Pakistan, but the international environment is sufficiently changed and the domestic employment crisis sufficiently serious, to pursue the matter with new vigour.
Samir Radwan
Director, Development Policies Department
This study on employment in Pakistan was conducted in 1997-98. Its objective has been to analyse trends in the labour market in the country, and develop possible elements of an employment strategy during Pakistan's Ninth Five Year Plan period ( 1997-98 to 2001-02).
A. Recent Changes in the macro environment.
Since the completion of this work in 1998, some factors have entered into Pakistan's planning process that have affected it significantly. The first is a rapidly changing macro environment. The second are the results of the 1998 Population Census. The report argues that these factors heighten the need for an employment strategy being proposed in this document. Its should however be finessed towards the transforming macro environment, and our better post-census perception of it.
Pakistan's macroeconomic performance, while showing indications of improvement at the time of the 1998 budget, has since been subject to fiscal stress. The central point that is worth making is that over the fiscal year 1997/98, growth had picked up in Pakistan to the five per cent level, while the IMF stipulated targets in macro fundamentals were met to the satisfaction of the evaluating mission to Pakistan at the time. However, domestic fiscal policy, perhaps in anticipation of an impending sanctions based global environment due to the nuclearisation of the country, (and possible investment behaviour based on this global environment), moved towards regulation in a number of areas (1). A major result of this macro transformation had been the cessation of the Enhanced Structural Adjustment Facility (ESAF) and Extended Fund Facility (EFF) agreements signed with the IMF, and a renegotiated new loan.
Although the external environment is now more stable, and donor links have been reestablished, changes in the macro environment need to be tracked as they can have critical implications for growth and employment in Pakistan. Clearly this issue needs to be assessed in future work. It is also critical to recognise the status of the specific macroeconomic policy environment for the question of whether growth is employment-friendly or not. Since the link between macropolicies and employment is generally mediated by growth, this provides an overarching context to understanding different phases of (employment focussed) economic development in Pakistan. The report devotes a separate Appendix to this issue. The reason to keep this discussion separate from the main body of the report is two fold. First that the assessment of the linkage between macropolicies and employment in this report is itself limited to an historical overview. Second, that it is the last phase of macropolicies in operation in Pakistan that are of greater importance to an employment strategy, than what can be argued for earlier phases.
B. The argument
The position of this study on employment and its determinants in Pakistan is the following. Wage employment in Pakistan is approximately one half of all employment, its proportion is high for a low income country. On the other hand self-employment comprises the other half of employment (Chapter 1). The share of the labour force in the estimated population for 1997, according to pre-1998 Census projections( on which this study has been based, and on which more below) is low. It was 37 million from a universe 135 million, which suggests a low participation rate. In fact the labour force participation rate (LFPR) in Pakistan has been decreasing over time. It is now down to around 27 per cent, therefore implying a high and increasing dependency ratio for the rest of the population. While we can explain to some extent, the drop in the LFPR to be due to increasing school enrollments (which is therefore in some respect desirable), there is clearly a possible trade off involved due the narrowing of the earning base in the country, which can put downward pressures on per capita income. This can have implications for poverty and living conditions of households. We actually find a mixed result on poverty in Pakistan, but it is one which is not inconsistent with the hypothesis of a downward pressure on living standards due to increasing dependence during the growth process.
The next question the report asks is what determines the low labour force participation and its income? The labour force, by definition, comprises of the employed and the unemployed, given at around 35 million and 2 million respectively. Before summarising the answer to this question, some qualifiers need to be made. There has been considerable debate and now recognition, in Pakistan, that women's income enhancing labour, whether in self-employment or in the wage labour market has been considerably under-enumerated due to non-sampling errors like male social biases, that render women's labour "invisible". This feature also reduces the overall estimates of the size of the employed. The second qualification is that under enumeration of the unemployed obtains because the survey- definition used in Pakistan requires them to be available for work in the reference week. Moreover it is to be remembered that these qualifiers obtain in the context of a labour market where formal institutions for employment are eclipsed by informal institutions. So, our estimates of both the employed and the unemployed, and the size of the total labour force are probably low.
This study assumes that given the labour force size, perhaps in this case based on consistent underestimation, what will determine the level of employment in the economy is firstly growth itself. On the other hand what determines the earnings of the employed are the wage rate and returns to self-employment in different sectors of the economy. In order to investigate the relationship between employment and growth, this study uses two analytical devices. One is a standard elasticity coefficient for employment with respect to growth. Problematic as the specific magnitude of the elasticity measure is, in the view taken here, it is considered at least qualitatively and directionally indicative of the employment-output linkage. This of course gives a range of values indicating say high, medium, or low responsiveness of employment to growth. The second device dis-aggregates change in output over time into (i) an output change caused by employment change; (ii) an output change that is caused by productivity changes; and (iii) an output change that is caused by a multiple effect of both employment and productivity changes. We refer to this as the decomposition exercise. The results of both exercises ( elasticity and decomposition) are used as summary perspective devices in the overview as well as sectoral chapters.
Chapter 2 of this study shows that over the last three decades, Pakistan's growth has been high for a low income country, around six per cent on trend till 1992/93, with a two per cent variation. However, against this high GDP growth, employment growth has been much lower, at two per cent on trend. As a result the elasticity of employment has remained medium to low, ranging between 0.6 and 0.3. The periodicity of this growth of output and employment is revealing, if we abstract for the moment from the issue of the particular periodisation chosen.
Output growth in Pakistan was low over the 1970s at under five per cent, it increased over the 1980s to above six per cent, and has dropped again over the 1990s to near about four per cent. Employment growth interestingly follows an opposite pattern, being at its highest at three per cent over the 1970s and lapsing to two per cent over the 1980s and the 1990s. Does this constitute some prima facie evidence for a trade off between employment and output growth? A further subdivision of these periods shows that there was a period of high employment in the 1970s, stable employment till the mid-1980s and a total break between growth and employment subsequently. The aforementioned decomposition exercise adds an insight to this pattern. Growth of output was based predominantly on an employment effect in the 1970s, on a productivity effect in the 1980s, and a mixed productivity and employment effect in the 1990s. So the answer to the trade off question posed above is not straight forward, and apart from having a source of output growth dimension ( productivity or employment) the answer is likely to have a sectoral dimension as well.
We know that generally low elasticities accompanying growth in Pakistan tend to conceal the fact that the source of growth (employment or productivity) itself has been varying over time. This is an important point to note. It is this changing relationship between growth of output and employment over time that is traced to its major sectoral components in the rest of the report.
In output growth, manufacturing is the leading sector (and agriculture contributes progressively less over time), over the low growth period of the 1970s, the high growth period of the 1980s, and the lower growth period of the 1990s. In employment growth, during the low growth period of the 1970s manufacturing leads, and with the relapsed of growth of the 1990s, employment growth in manufacturing turns negative.
The explanation for the changing overall relationship between output and employment growth over time then finds a possible sectoral reason. It has been said that employment growth was high over the 1970s and then stable till the mid 1980s, after which it de-linked from output growth. Manufacturing contributed relatively more to employment growth than agriculture in the 1970s, relatively less than agriculture in the 1980s, and negatively in the 1990s. So the overall break between output growth and employment seems to be more associated with the manufacturing sector. It is also true that we find some evidence to suggest that agriculture and construction are becoming possible refuge sectors for labour after this de-linking of employment and output growth. This explanation is further enhanced when we examine output and employment growth relationships at the sectoral level through the same two analytical devices of sectoral elasticities and the decomposition exercises. Chapters 3 to 5 examine sectors.
Chapter 3 on agriculture shows that output growth and employment growth (and therefore, employment elasticity) were stable in the 1970s till the mid-1980s, barring states of nature based interventions. Therefore, output and employment growth are connected till the mid-1980s. After the mid-1980s, output and employment growth became volatile. The elasticity of employment in agriculture is thus positive and stable till the mid-1980s then subsequently starts fluctuating. The decomposition exercise for agriculture shows that agricultural output growth was driven predominantly by the employment effect in the 1970s, largely by the productivity effect in the 1980s and then again largely by the employment effect in the 1990s. The report does pursue alternative explanations for the contribution of agriculture to the fluctuations in employment from the mid-1980s onwards, in terms of the increasing sensitivity of a post-High Yielding Variety (HYV) agriculture to states of nature, and the casualisation of the labour force based on increasing mechanisation.
In contrast, to agriculture, Chapter 4 shows that in the case of manufacturing, employment growth was stable and high over the 1970s, when in fact output growth was low. In the 1980s, output growth in manufacturing rose, but employment growth dropped to a stable low, the aforementioned critical disjuncture become evident by 1980. By the mid 1980s, output and employment growth both become unstable. Elasticity of employment in manufacturing, barring a peak in the mid-1970s remains at a continuously low level. The decomposition exercise for the manufacturing sector highlights this break between employment and output growth after the 1970s. Over the 1970s output growth in manufacturing is driven equally by the employment effect and the productivity effect. Over the 1980s the employment effect becomes negligible, while over the 1990s it turns negative.
So, the disjuncture between employment and output growth rates in manufacturing reflect and track the economy-wide break between employment and output growth rates better. Low output growth accompanies high employment in the aggregate over the 1970s, high output growth accompanies low but stable employment growth till the mid-1980s, after which employment and output growth become volatile. It is fairly clear that both agriculture and manufacturing contribute to output through a predominant employment effect in the 1970s, but for the 1980s and 1990s, only agriculture contributes through an employment effect, while manufacturing's employment growth breaks away from its growth of output.
Having focussed on the manufacturing sector, we need to trace the causality further. In other words we need to determine the causes for this break between employment and output growth in manufacturing itself. The first point to note here is that while the share of manufacturing in GDP has increased over time from 15 per cent to 18 per cent, its share in employment has declined from 14 per cent to 10 per cent. To explain why the manufacturing sector has become entirely labour productivity driven rather than in some measure employment-driven, we have to recognise some structural features of the sector. The critical feature of manufacturing in Pakistan, as in many developing economies is its duality. The large scale (LS) sector and a small scale (SS) sector are two classifications of manufacturing and the cutoff point between the two has been defined variously over time by the government of Pakistan (GOP) (2). The LS sector (examined in Chapter 4), produces two-thirds of the value added in manufacturing, but employs only 17 pre cent of the approximately 4 million labour force in manufacturing. The SS sector (examined in Chapter 5), produces one third of the value added in manufacturing, but employs 83 per cent of the labour. So the LS sector dominates output, while the SS sector dominates employment. Further, the output and employment shares of the LS and SS sectors have remained fairly stable over the last two decades.
The differential between the two manufacturing sectors, LS and SS, is in labour productivity; and this in turn is reflected in a differential in their capital-labour ratios. The LS sector has a higher capital labour ratio by a factor of 17, and therefore a much higher labour productivity. It also has higher wage rates and greater job security, and it is argued over the course of Chapters 3 and 4, that LS growth rates are higher growth rates than those in the SS sector. Other production characteristics of the SS sector are detailed in Chapter 5, these are fragile vending linkages to other parts of the economy and negligible formal credit. The typical SS unit is based on self employment and family labour; with earnings approximating the LS sector's wage rate, with low vintage, and limited vertical transition out of the sector. These characteristics of the SS sector, while not completely homogenising potentially render a large part of it, a refuge labour sector in Pakistan.
The following question then is, if the SS sector has in large part been or become moribund with all its disincentives, why has the LS sector failed to expand its employment levels significantly over time, say between 1980 and 1995, when employment grew at 2.3 per cent, employment in the LS sector grew at under 1.8 per cent. The first element in our answer is an increase in capital productivity in large scale manufacturing over time (3). Furthermore, Chapter 4 also shows for the LS sector, that while this increase in capital productivity was positively correlated to labour productivity, it was significantly negatively correlated to employment. As a result the wage share in output dropped from 0.4 to 0.29. This is accompanied by the fact, as Chapter 1 shows, that the real wage in Pakistani industry barely creeps up over the last two decades. It is worth pointing out that this pattern of growth in the LS sector was enabled by a large increase in investment in manufacturing, and liberal financial sector lending. So over the period 1982/83 to 1992/93, private investment in manufacturing increased from 1.7 per cent of GDP to 4.2 per cent. In other words, the pattern of investment and growth in manufacturing, over the 1980s, was capital augmenting and labour displacing, and it increased profitability and reduced the wage share in output. The evidence tends to favour this view (4).
The second element in our answer to the disjuncture between growth of output and employment in the LS sector lies in the creation of excess capacity in the 1990s. Private investment in manufacturing halves after 1993/94 from 4.2 per cent of GDP to 2.2 per cent with a declining trend, as Chapter 4 shows. This results in a declining trend in capacity utilisation in the four major industries of the LS sector, spinning, weaving, cement and sugar. The economic position taken here is that this downturn in investment, capacity utilisation, and growth in manufacturing is caused in large part by a failure in demand, significantly associated with the deflationary policies pursued by GOP in accordance with the Structural Adjustment (SA) programme signed with the IMF, in successive protocols since the 1988 .
What probably happened was a coincidence of events. Growth in Pakistan has always been susceptible to states of nature caused downs in agriculture, given its preponderant reliance on the textile industry and export of cotton and its products. The dips in GDP in each decade, examined in Chapter 2, are related to flooding in agriculture in the early 1970s and in the early 1980s, and a pest attack on cotton in 1992/93. However in the 1970s and the 1980s, the economy recovered with a one to two-year lag, returning to trend growth. In the 1990s, however, this return to trend growth has arguably been inhibited by the simultaneity of deflationary SA policies with the massive decline in cotton output for two years running. Just when the state of nature caused slump needed a reflationary policy, the SA programme veered the economic managers towards policies effectively increasing the incidence of general taxation and a decline in subsidies. This was affected through the imposition of a Value Added Tax (VAT) and successive annual increases in its rate and coverage; and through a reduction in the budgetary deficit especially affecting agricultural subsidies and therefore production and recovery. This was also manifested in a decreased purchasing power and effective demand.
In summary, we have found that the relationship between growth in employment and output is strong over the 1970s, and stable over the first half of the 1980s, but becomes negative subsequently. Manufacturing is the leading sector in growth of output and employment in the 1970s, but contributes little to employment in the 1990s. The manufacturing sector is seen to be characterised by a dualism, with an LS sector dominating output, and an SS sector dominating employment over time. Given low capital labour ratios in the SS sector and all the disadvantages in production conditions that this entails, the central question that emerged was why the relatively more advantaged LS sector had constrained its employment levels over time, especially over the 1980s and the 1990s. Two elements were identified to answer this question. One, that over the 1980s, the LS sector in Pakistan followed an internationally established pattern of capital augmenting, labour substituting growth, increasing profit shares and reducing the wage share. Two, that over the 1990s, deflationary SA policies dampened recovery from a state of nature caused blip in growth, especially for manufacturing, constraining capacity utilisation and employment further.
An employment strategy for Pakistan must then seek to redress this disjuncture between growth of output and employment, caused in large part by the LS sector in manufacturing. This is what is proposed in this report. The key constraints are an overwhelming capital augmenting growth path, and deflationary SA policies. This brings us back full circle to the starting point this argument. With an impending, and even larger, SA programme in the offing, Pakistan faces the prospect of possibly low growth and low rate of recovery from a fiscal crisis as in 1992/93, and possibly employment-less growth as in the 1980s. This makes a planned employment strategy for the NFYP period difficult but imperative. On the other hand, the more recent changes in the thinking of Bretton Woods institutions, with respect to the recognition of social outfalls of adjustment programmes, allows Pakistan, as indeed other developing countries, a better context in which to make a case for a macro and social policy framework which is consistent with a positive employment strategy.
A well-known shortcoming of analysis on labour markets in Pakistan is the internal limitations of the data produced on population, labour force and employment. Coverage of the labour force surveys (LFS) is known to be low, with the result that both participation rates and employment elasticities based on LFS data are thought to be underestimated. The population census scheduled for 1991 had been delayed, leaving a gap of over seventeen years since the previous census was taken. The full results of the Population Census of 1998 were still unavailable at the time of the writing of this report. The absence of regular and better data confines us to work, with due care, with the data available (Annex 1). There is, however, a distinction (if not a choice) here which needs to be made. In view of the weight of empirical evidence that does (or does not) exist, there are two courses open for further investigation. The first is to accept the trends from large-scale data as a starting point and then investigate the issues and problems this data throws up at the disaggregated level, in order to qualify the existing trends - both sectorally and with respect to special categories/groups of labour that require focus. The second course is to reject the census and survey data on the grounds that the true labour supply is understated (1) and proceed with analyses supported by the micro evidence that can be marshalled, without regard for the larger picture.
The view taken in this report is that there is a need for greater caution in rejecting the burden of numerous surveys, irregular and imperfect as they may be and even if they yield apparent inconsistencies in their results, as surveys anywhere are apt to do. The size of the information that does exist in Pakistan is substantial. In order to judge the validity of anomalies in the larger data sets, what is ideally needed is a specific survey, of equal or greater coverage, which provides empirical proof of the alleged bias in previous census and survey data. There is both an absence and a critical need for information on labour market indicators in Pakistan. Given the lack of a perfect series or a second-best one, we will work with what is available, using annual data from labour force surveys. These are subject to much fluctuation, suggesting a weakness in data source. Our emphasis is therefore more on trends than on explaining annual fluctuations, where they seem inexplicable.
The earliest and, probably, still the more reliable data on population and labour force are provided by the population censuses of 1951, 1961, 1972, and 1981 (2) (See tables A1.1, A1.2 and A1.3). In lieu of the much delayed census for 1991, we have used population and labour force data from the Demographic Survey of 1991 and the annual labour force surveys carried out since. These are used to assess the employment situation, especially after 1981 (Table A1.2). (3)
The labour force (around 37 million in 1997) is not a very large proportion of the population (135 million in 1997) in Pakistan. Its is estimated to be around 28-29 per cent. Population growth had been increasing over the years, up to and including 1970s. Although the last census was carried out in 1980, projections on the population growth rate suggest that it is likely to have fallen and has probably stabilized now to just under 3 per cent (4).
Labour force growth rates on the other hand were higher than that of population in 1970s (figure 1.1 and Table A1.1) but have been generally lower after the 1970s. Although the labour force survey figures for the 1990s can also be interpreted to suggest a decline in labour force growth on trend (5), there are problems in comparing the labour force trends before and after 1990-91 (6). There is however a clear decline between the two definitionaly comparable periods of the 1970s and 1980s.
This means that labour force growth, which is significantly lower than population growth, is likely to be falling while the growth of population is likely to be stabilizing.
If serious data problems are excluded as a cause, then one of the main reasons for falling labour force growth rates is reflected in the decline in participation rates due to withdrawal of sections of persons of working age from the labour force. Educational enrolments for an increasingly younger population of working age mean that the labour force growth will begin to rise in the near future. This has implications for trends on unemployment and underemployment in the present situation. These are likely to be magnified with an increase the growth rate of the labour force in the future.
Definitions of labour utilization
The "economically active population" comprises persons of either sex who furnish the supply of labour for the production of goods and services, during a specified time reference period. But not all these persons are currently active (or in the labour force) at any particular point of time. In Pakistan, the currently active population or labour force comprises of all persons ten years of age and over who fulfil the requirements for inclusion among employed or unemployed, defined as follows, during the reference period i.e. one week preceding the date of interview. (7)
The self-employed amongst the employed
Employment is divided into two categories in Pakistan: the self-employed and the hired. The self-employed are only marginally less in number than the hired within the labour force. During 1983-95, the self-employed, on an average, constituted 44 per cent of the employed work force: 46 per cent in the rural areas and 37 per cent in the urban areas. The percentage of the self-employed increased from 40.3 in 1982-83 to 47.9 in 1987-88 but was at 42.3 in 1994-95. This percentage increased in the rural areas and declined in urban areas during the period 1982-83 and 1994-95 (table A10).
In 1987-88, the agricultural sector accounted for 53.6 per cent of the self-employed, followed by trade (16.8 percent) and manufacturing (10.9 percent). In the rural areas, agriculture accounted for 66.6 per cent of the self-employed followed by 9.8 per cent in trade. In the urban areas, the wholesale and retail trade sector accounted for the major share of 41.8 per cent (Table A1.11).
Another distinct feature of the self-employed is that compared to the national average, the share of self-employed who worked full time was higher (except for women in rural areas). Moreover, the level of education of the self-employed was also lower than the national average (Table A1.12).
The self-employed are sizeable, significantly agrarian, work long hours and are relatively less literate than their counterparts in employment. It is important to bear in mind that self-employment in Pakistan is predominantly in agriculture and more symptomatic of poverty than prosperity.
Underemployment
The underemployment rate is estimated as a ratio of employed who worked less than 35 hours a week to the total labour force (Table A1.13). There were nearly 4 million persons underemployed in Pakistan in 1994-95.
The proportion of employed persons who worked for less than 35 hours, which had declined from 14 per cent of the employed work force in 1982-83 to 11.1 per cent in 1987-88, was at 12.2 per cent in 1994-95. Underemployment was much more prevalent in rural than in urban areas. In 1987-88, it was 2.8 times that of urban areas. In 1993-94 it was almost double. Unpaid family helpers constituted about half of the underemployed, followed by the self-employed. This is true in the aggregate as well as in the rural areas. Underemployment among employers was almost negligible.
Within rural areas the share of underemployed declined from 16.3 per cent in 1982-83 to 13.4 per cent in 1987-88, but was at 14 per cent in 1994-95. In contrast, the share of underemployed in urban areas declined from 7 per cent in 1982-83 to 4.8 per cent in 1987-88, but was at 7.1 per cent in 1994-95 (Table A1.14).
It is also interesting note that in the 1993-94 LFS (when this question was asked) a four-fifths majority of those who were underemployed, considered their working hours as "normal". This implies a "normality" of being permanently underemployed, in an "enclave" of the labour market where most of the underemployed are not so out of choice (only 2.6 per cent in this group describe their underemployment as voluntary) and where the severity of underemployment is unlikely to decrease.
Unemployment
In a developing economy characterized by significant poverty, minimal social insurance and welfare mechanisms, unemployment is probably the least appropriate indicator of labour market conditions. Most people of working age, who are not in full-time education, need to work to survive. With this proviso in mind, we will examine unemployment data.
First, it is necessary to clarify a definitional problem. Before 1990-91, the LFS recorded those persons as unemployed who were without work (i.e. were not in paid employment or self-employment) and were both currently available and seeking work during the reference period of one week. Those who may have been available for work but for some reason were not seeking it (i.e. not taken specific steps to seek paid or self-employment) were not recorded as unemployed. The unemployment rates prior to this period were therefore recorded at a low figure of about 3 per cent. From 1990-91 onwards, the condition of being available for work was sufficient to be recorded as unemployed. This resulted in the unemployment rate in 1990-91 jumping to 6.3 from 3.1 per cent in 1989-90 (Table A1. 15) and implies that comparisons on unemployment (8) are not legitimate across the year 1990-91.
The number of unemployed persons was two million in 1990-91. This figure was one million in 1989-90 and much of the increase occurred because of the change in definition of the term. The overall unemployment rate declined in 1991-92 and 1992-93 but it went up in 1993-94 and remained at the same level in 1994-95 and declined again afterwards. There is no real trend which is discernible (9). For general policy purposes, it may not be wrong to assume that currently about two million people are unemployed in Pakistan. According to data, 63 per cent are men and 37 per cent women. Of the total unemployed in 1994-95, about 60 per cent were in rural areas. However, the unemployment rate (ratio of unemployed persons 10 years of age and over to labour force) was much higher in urban areas (6.51 per cent in urban as against 4.22 per cent in rural areas). Available figures for 1984-85 to 1987-88 show that the severity of unemployment as judged by its duration as well as the percentage of unemployed with previous experience has been increasing (Table A1.16).
Unemployment rates are higher both for older and younger workers. Whereas the high unemployment rates for the 10-14 age group may be attributed to restrictions on (child) labour below the age of 14, the high rates in the 60 plus age group is explained by the non-availability of jobs suiting older workers. (10) Within the unemployed, the share of literates has increased on trend from 42.8 per cent to 48.8 per cent (Table A1.17).
At the aggregate level the unemployment rate has averaged around 5.0 per cent a year during 1990-91 to 1994-95. Of two million unemployed in Pakistan, nearly half of them are literate and the severity of their unemployment is likely to be increasing.
Employment and labour force: An overview
Employment growth rates generally increased in the 1970s, and from the end of the 1970s declined on trend till the late 1980s. In the 1990s, although it is difficult to make a comparative statement, employment growth is falling on trend. (11) The variability in growth rates is partly explained by definitional changes. (12) In sum:
Given that labour force growth is expected to increase in the future, those employment-related problems that exist now, under falling labour force growth conditions, are likely to worsen.
We now make some further observations on the data available.
In absolute numbers, the labour force in Pakistan has increased from 18.1 million to 35.2 million between 1972 and 1995 (Table A1.3). The male labour force has increased from 16.7 million to 30.5 million and the female labour force from 1.4 million to 4.7 million. Compared to an average annual increase of 2.7 per cent in the total labour force between 1972-95, the urban labour force increased by 3.5 per cent per year, while the rural labour force increased by 2.5 per cent per year in the same period. As a result, the share of the urban labour force in the total labour force increased from 23.3 per cent in 1971-72 to 27.3 per cent in 1994-95. This difference in growth rates was largely the result of rural-urban migration. The growth of the rural labour force in the late 1970s and the first half of the 1980s was also affected (restricted) by emigration, which is much less important today.
Traditionally, and certainly since 1975 onwards the growth of the urban labour force has been faster than that of the rural labour force. This was to be expected. However, the reversal of trends in growth rates in the 1990s (rural labour force growth rates being higher than urban ones) (14) suggest that, if no serious data problem exists, there may be an emerging barrier on internal rural urban migration (15) (Table A1.3). This issue has implications for the size of the labour force in the rural areas, particularly in agriculture, and needs to be explored.
The size of the labour force is determined by the size of the working-age population and its activity or participation rates (16), adjusted for migration into and out of the country. This population has already been born; the demographic changes which occur today will affect labour supply with a time lag of about 10 years. The "activity rate" is therefore the key determinant of the currently active population or labour supply.
We need to explain why the participation rates of labour force are declining in Pakistan. The LFPR peaked to 31 per cent in 1978-79 and declined to 27.5 per cent in 1994-95. The falling labour force participation rate has clearly restrained the growth of the domestic labour force (Table A1.4). (17) It is worth noting that rural participation rates are generally higher than urban ones, another reason why when labour force growth declines it does so less in rural areas. This is also reflected in the fact that while the rates of male participation in the labour force do not vary so much across the rural and urban divide, and they do so less over time, the gap between rural and urban participation rates for women is significant and persists over time. The greater participation in rural areas is due to a greater inclusion of women in what constitutes the labour force, in spite of the reporting biases on women's household activities in both rural and urban areas. This feature comes into sharp relief when we consider the size of the excluded labour force in Pakistan.
We discuss below three tendencies for changes in the LFPRs in Pakistan whose net impact is clearly a decline. These pertain to education, emigration and gender.
Education and the labour force
High population growth in Pakistan has also been associated with an increasing proportion of population in the lowest age cohorts (Table A1.5). The proportion of population in the 0-14 years age group increased from 44.5 per cent in 1981 to 45.8 per cent in 1988 and to 46.3 per cent in 1995, thus raising dependency ratios and depressing labour force participation rates. In part, age compositions do explain why the overall participation rates are low. The 10-14 years age group show a decline in participation rates, suggesting a withdrawal of the young (who are counted as economically active) from the labour force, possibly due to increased school attendance. (18) The decline in participation rates, in the 15-19 years age group may be similarly attributed to increasing college enrolment.
Emigration and the labour force
Increasing emigration from the mid-1970s to the mid-1980s had constrained growth in domestic labour. According to the report of the National Manpower Commission (Government of Pakistan, 1989, p. 101), during 1978-83, almost 25 per cent of the incremental labour force found employment abroad. This helped in sustaining employment growth at 2.8 per cent. However, return migration in 1982-87 dominated the stream of migration and may have played some part in accentuating pressures on the domestic employment situation (Table A1.6). For the more recent sub-period there is evidence of net out-migration. The net outflow in 1993-95 period is, however, a much lower percentage of the labour force, in comparison with the past peak of the 1980s. The size of the labour force today is not subject to pressures return migration and is unlikely to be seriously constrained by out-migration.
Gender and the labour force
This is an area which requires serious analytical and empirical disentangling.
Due to the way the term 'labour force' is defined, a number of persons, particularly women engaged in household duties, are not included in the term 'population currently active' or labour force (table A7). This reduces the number of labour force as well as the participation rate, a shortcoming that was partly made up for in Pakistan by adopting a new definition in the 1990-91 LFS, in which women were identified as employed if they spent time on any of the 14 specified agricultural and non-agricultural activities defined in the Intentional Standard Industrial Classification, (ISIC). (19) This partly redresses the definitional bias. However, it is interesting to note that in 1994-95 about 59 per cent of the economically active population were excluded from the labour force. Of these, about 27 per cent were men and 73 per cent were women. Clearly, the bulk of the 'excluded labour force' are women. (20)
An obvious reason for this in Pakistan is that social perceptions restrict the 'admission' of the existence of female work, by a not too insignificant section of the population. This implies a reporting bias due to which female participation rates and employment can be seriously under-reported. Secondly, in Pakistan, much of the employment data, irrespective of its source, is collected by male investigators interviewing male respondents. (21) Apart from the degree of authenticity of responses, variations in reported employment may also result from variations in the education, training or supervision of investigators; the periods chosen for the survey; and the language used in the questionnaire. (22) These elements vary from survey to survey and can lead to serious inconsistencies in the published data.
The definitional changes in LFSs in 1990-91 (apart from increasing the proportion of unemployed in the labour force) improved women's participation rates in general. This may be more an effect of a categorical change and less of an actual improvement in women participation rates, which are still very low. This low rate accounts for the low overall level of participation of the labour force shows up as a trend decline in participation rates and the labour force growth over the years. For the 1990s the improvement in coverage of working women is offset by the withdrawal of the young from the labour force.
The composition of labour force by level of education has witnessed changes in Pakistan (Table A1.8). With the improvement in the literacy rate, the proportion of the literate labour force has increased from 39.9 per cent in 1990-91 to 43.6 per cent in 1994-95. While the share of literate men in the labour force increased from 37.3 per cent to 40.1 percent, that of women increased only marginally from 2.57 per cent to 2.67 per cent. (23) Clearly, if one criterion for employability is literacy, then an explanation of low participation rates of women is low literacy, which needs to be a policy focus.
The literacy rate of the labour force has risen from 35.3 per cent in 1987-88 to 43.2 per cent in 1994-95. Over this period, the share of degree holders has shown the sharpest rise (from 1.8 per cent to 2.9 per cent), followed by that of matriculates (from 9.4 per cent to 13.2 per cent). The composition of employed work force by the level of education has also undergone significant changes (Table A1.9). (24)
Within the literate labour force, employment growth rates have been positive for all levels of education except for the holders of degrees in medicine and those with no formal education (25). The proportion of literates in the labour force is also increasing albeit slowly.
Although literacy is increasing 56 per cent of the labour force are still illiterate and about another 10 per cent literate only at pre-matriculate levels or with no formal education. Efforts to introduce new technologies and management techniques in the short-run, essential for the improvement of productivity and efficiency, are likely to be constrained.
Labour force expectations and the structure of the job market
Although it may be true that economic growth itself is fundamental as a final adjuster of the labour market in the medium run, it is also arguable that, at least in part, mechanisms of adjustment depend upon the perceptions of those who bear the burden of adjustment in the labour market, namely the unemployed and the underemployed.
A question was put in the LFS of 1993-94 to those who were not working in the reference week regarding work expectations. These expectations are indicative of a mismatch with prevailing sectoral, as well as locational, conditions of employment. This is an interesting finding, which needs some elaboration. In the LFS of 1993-94, nearly two-thirds of the unemployed expressed the wish for employment with government, a proportion which, if realized, would be more than twice the current governmental share in employment. Nearly one-fourth of the unemployed preferred clerical and related jobs. This is mainly the response of the matriculates and intermediates. On the other hand, only 1 per cent of the unemployed said that they would like to work in the agricultural sector while nearly half of Pakistan's employed are still in the occupational category of agriculturalists and forestry workers. Occupational preferences of the unemployed do partly reflect their educational background, but the dissonance which is being reflected here is really between the development of an educational system and the production structure and needs of the economy.
Orientating a future labour force (which will increase in growth and literacy) towards the types of activities the future holds should be an essential part of the training and skilling efforts being made today. This is an important challenge for the future of both the labour force and the country.
Pakistan had a population of around 128.3 million in 1994-95 of whom 85.4 million were above the age of 10. The labour force size is 35.2 million (29.3 employed; 4.0 underemployed and 2 unemployed). Those who were outside the labour force, excluding students (17.4 million) were around 32.8 million, 30.1 million of whom were women. The growth of the labour force has been lower than that of population on trend in the past. Urban labour forces have also been growing faster than rural ones due to rural-urban migration, although there are some indicators in the 1990s that internal migration may be halting and trends could be reversing.
The rate of growth of the labour force in Pakistan has been falling in the past but it is expected to rise in the near future. Unemployment and underemployment trends suggest a worsening or certainly no improvement in the recent past. This means that the problems associated with those who are not fully employed is likely to increase when labour force growth picks up. The labour force is also slowly becoming more literate in Pakistan.
Apart from the matters pertaining to restrictions in the growth of population, there are three areas which require focusing from an employment perspective. These relate to data, participation rates and training.
Data
The coverage of the LFS must improve. This is with respect to the excluded sections of the adult population, which are dominated by women. Moreover, definitions need to further improve to include home workers, as well as other categories of persons who may be excluded for non-technical reasons. The Population Census must be conducted without delay. It is the fundamental basis of any serious policy work, quite apart from being a necessary cross-check on the Labour Force Surveys. A country the size of Pakistan cannot afford to avoid a Population Census for seventeen years, no matter what the domestic political fall out of this may be. Consistency exercises must continue to be in built into data collection, as was the case in the 1990-91, whenever definitional changes take place, so that trends can be readjusted. A consistent series of best-possible estimates of all labour market indicators in Pakistan needs to be built, which adjusts and reconciles the data available from different sources.
Participation Rates
Clearly, participation rates have been very low and declining in Pakistan. These need to be improved for the productive absorption of the labour force in the economy. The issue of female participation rates is a subset of the general issue. The reason for the increase in the low participation rates for women is not necessarily only because more women are working but because definitions have improved. The causes for the low participation rates as well as high incidence of women in those excluded from the labour force need to be ascertained carefully. The extent to which this is due to social factors, definitional problems, or coverage need to be distinguished empirically. On a general plane, the low participation rates needs to be improved to effect a more productive use of labour in the economy.
Training and skills
There must be a conscious attempt made to bring the expectations of the literate labour force which is either not in employment or likely to come into the labour force in the future, into line with actual possibilities in the job market. This should be done both at an educational level as well as at the level of training programmes.
Implications for a strategy
From the perspective of a strategy, two issues are important:
The magnitude of the employment problem facing Pakistan in the future is considerable. Even if the rate of growth of the labour force is likely to pick up in the near future, the deteriorating employment situation is likely to worsen more than present trends anticipate.
There are two related matters for concern: the gap between population and the labour force and a rising labour force in comparison to the "fully employed" within it. An increasing gap between population and labour force growth for a given output implies the increasing dependency of a rapidly growing part of the population on a slowly rising labour force. Under these same conditions, the increase in the gap between the fully employed and the labour force implies constraints on the earning capacity of the labour force itself. This has direct and demonstrable implications for poverty. (1)
One goal of a future employment strategy is therefore to reduce the growth of population. The other, and what we are concerned with in this report, is to productively increase the participation of the labour force in the economy. In doing so, the aim is to improve aggregate employment and attack the twin problems of unemployment and underemployment head on. The next sections take a more detailed view of the employment situation.
Given the critical situation described above, employment strategies need to be devised in order to face the employment challenge. The absorption of a larger percentage of population in gainful work, as well as the reduction of the under employed and unemployed, are important goals.
It would, however, be hazardous to proceed as if all that needs doing is to direct efforts towards an expansion of sectors with higher employment elasticities and wait for employment to happen. The precarious nature of estimating elasticities of employment in Pakistan has been discussed elsewhere (2). For the purposes of the present analysis, our position is that elasticity estimates, though useful in the generation of ball-park projections, are on their own, insufficient for planning an employment strategy. Moreover, the estimates tend to become more inconclusive the better the methodology used. Elasticities have generally been low in Pakistan and they have been lower in the high growth period of the 1980s
than they were in the relatively lower growth periods of the 1970s and 1990s (3). The increasing trend in employment elasticities are however, is not a simple index of ascertaining the "labour absorption capacity"of an economy. We need to look at other indicators implicit in the elasticity measure, in order to make some judgements on the quality of the labour absorption which the elasticity measure signifies.
Although sector-specific analyses are necessary for producing the details of an employment strategy that can lead to concrete policies, an aggregative analysis of the sort attempted in this chapter can still provide a plausible a sectoral prioritization within an economy-wide framework. The following sections examine the sectoral trends in employment in Pakistan before proceeding to study the indicators of employment implicit in an elasticity measure: growth in output, employment and labour productivity.
As far as sectoral distribution of output shares are concerned, in the long run agriculture's share declined from 39 per cent in 1971 to 24.3 per cent in 1997 (Table A2.2). GDP shares of manufacturing have increased on trend; they were 14.16 per cent in 1971 and 17.92 per cent in 1997. There are small increases in other sectors' GDP shares. The decline has essentially been for agriculture and the gain for most of the other major sectors. This is to be expected in the process of development.
Sectoral shifts have also taken place in employment shares over time. Once again for agriculture there has been a long-run trend decline but from the mid-1980s the share has been fluctuating between the 45 per cent and 55 per cent mark (Table A2.3). It can be argued that although the share of employment has declined for agriculture, over the longer-run period, it has been fluctuating within a range in the last decade. On the other hand, contrary to expectation, there has been a declining trend in the share of manufacturing in employment from 1969-70 (15.57 per cent) to 1993-94 (10.12 per cent), although employment shares on average rose slightly from the 1970s to the 1980s and then declined in the 1990s. All other sectors which include construction, trade and the "other" activities category over a longer-run period, from say 1970-71 to 1994-95, have shown an increase in employment shares. It is therefore quite reasonable to argue that accompanying the longer run decline in employment in agriculture, say from 1969-70 onwards, there have been increases in employment shares of sectors other than manufacturing.
In the recent and shorter-run period of the 1990s, however, agriculture's fluctuating share of employment forms no trend (except that it breaks from the past trend of a declining share of employment), while the share of manufacturing (and mining) in employment has definitely fallen from 12.54 per cent in 1990-91 to 10.12 per cent in 1993-94.
Although we need to be aware of the standard limitations of the elasticity measure, it is still worth examining the trend in employment elasticities, and of output and employment growth for the economy as a whole over the years. This in conjunction with other measures can help identify general trends. Figure 2.1 below splits the employment elasticity measure into its two components: employment growth and output growth.
The economy-wide picture shows a few features clearly.
This leaves the matter of the recent volatility in employment growth, which appears seems to be a new structural feature in the Pakistan economy, as an issue that needs to be explained. Examining sectoral employment changes over time delineates the sources of this volatility.
One question that needs answering in an economy-wide context is: why has the structure of employment growth become so volatile? (4) As a first step, this requires one to explore sectoral trends in employment in more detail. The Figure below presents data on annual changes in the number of persons employed by sectors. (5)
Although the data presented are not free of problems, (6) they do show that:
The employment swings discussed above and the trends in elasticity observed are consistent with the decomposition of output growth presented for the 1970s, 1980s and 1990s decades in Table 2.1 below. Output growth decomposes into two contributing components: employment expansion and labour productivity and accordingly periodizes the Pakistan economy since the 1970s. (7)
Table 2.1: Decomposing output growth, productivity and employment effects
| dP . Nto / dY Productivity Effect |
dN . Pto/ dY Employment Effect |
dN . dP / dY Multiple Effect | |
| 1969/70 - 1980/81 | 0.292 | 0.615 | 0.118 |
| 1980/81-1990/92 | 0.578 | 0.273 | 0.148 |
| 1991/92 - 1994/95 | 0.400 | 0.571 | 0.028 |
| Source: National Income Accounts and Labour Force Survey (various years). | |||
The employment effect was strongest in the 1970s after which there was productivity- led growth in the 1980s. The high growth period (1980s) was least driven by employment expansion as a contributor. In the 1990s, there was a shift back towards a growth process which is partly based on increasing employment. Given that the 1990s are a relatively lower growth period, we need to probe further into the nature of this employment expansion. In the section following we examine the employment conditions which charactercized the changes, as reflected in wages and poverty.
Where organized labour is not the largest part of the hired labour force (8), and where a multiplicity of contractual arrangements exist, it is important to analyse not only real wages over time but poverty as well. Unfortunately, the database on wages is limited in Pakistan and poverty assessments not only differ in methodology but are also constrained by data sources that are not current. The importance of wages and poverty indicators necessitates an attempt at their assessment. We start by discussing wage rates (Table A2.10).
Real Wages
Wages rates are clearly more relevant to that part of the labour force which is hired, a figure previously mentioned as not much more than half of the employed labour force. About 56 per cent of the employed labour force is hired., the rest are self-employed. It is, however, not inconceivable that the poorer sections of the self-employed in Pakistan (particularly in agriculture) also offer themselves on the labour market as part-time casual workers. (9)
The percentage of hired labour which is permanent relative to those on casual contracts is likely to be small. Much of this organized labour force is either in salaried government jobs or in the organized private sector of the economy. A smaller proportion is likely to be in the organized industrial workforce. Consequently, the dominant form of hired labour in Pakistan, in an economy-wide sense, is casual workers, some who offer themselves on casual wages throughout the year, others who do casual work only as a part-time activity.
There are three wage rates available for the agricultural sector from the 1980s. One for each season (Rabi and Kharif), and a general casual wage rate. It can be argued that the dominant type of casual labour used in agriculture is season-specific and not of a general variety, the latter usually representing the casual labour available in rural areas. The wage rates for two seasons (Rabi and Kharif) in agriculture show an increase up to the end of 1980s and then a decline in the 1990s. The general and non-seasonally specific casual wage rates in agriculture also suggest an increasing trend till the end of 1980s and have then fluctuated. It is clear that overall real wage rates in agriculture increased in the 1980s and are witnessing a decline in the 1990s (Figure 2.3a).
Real wages of the organized industrial sector, have increased on trend from the 1970s, although there is no data available for recent years. As suggested earlier, this is likely to be a small proportion of the total employed labour force ( not exceeding 10 per cent). The category of skilled workers can be seen in real wages of masons: an increase in real wages in the 1970s, a decline in the early 1980s and an increase on trend from the late 1980s onwards (Figure 2.3b).
The real wage index of unskilled labourers is based on data of casual workers in construction. It can be taken as a proxy for non-agricultural urban casual workers, which are likely to dominate the urban small-scale and informal enterprise sector. This also shows a trend increase in the 1970s, and then a decline till the mid-1980s, which increases briefly during the late 1980s and then shows a slow decline through the 1990s. The pattern is clearer for real wages of unskilled casual labour an increase in the 1970s and a decline in the 1990s (Figure 2.3c).
Poverty Trends
Trends in poverty are difficult to estimate in Pakistan because research is often based on different poverty lines and a number of problems also exist with the types of data often available. There is, however, a need to view the trends in poverty in Pakistan, both rural and urban, from an employment perspective. (10) Malik (1988) has estimated poverty lines from 1963-64 to 1984-85 on a consistent methodology based on calorie norm. (11) Figure 2.4 shows that poverty increased in Pakistan during the 1960s (although urban poverty declined). It began declining during the 1970s and this decline continued up to the late 1980s. After the late 1980s (1987-88) there is every likelihood that poverty increased, at least up to 1990-91, after which it has either increased or remained the same. The 1990s trend is therefore likely to be one of increase. (12) It is true that the total poverty trend described above is generally reflected in the trends in rural poverty and not urban poverty. Urban poverty has shown significant declines when total poverty has increased (1960s and the 1990s). Urban poverty may have started declining again in the 1990s, while rural poverty is probably either not declining or is increasing. This is of immediate relevance for our impression of wages for the early 1990s and is quite consistent with the reversal of past urban-rural labour force growth trends in the 1990s.
Poverty, Wages and Growth
It is also useful to make some qualitative assessments about poverty and wages and growth taken together.
For the 1970s, it is fairly clear that poverty declines in both urban and rural areas were matched by a universal increase in all types of real wage rates for most of the period. Whereas GDP growth was not so high, the growth that took place was driven by employment expansion. It very clearly demonstrates the inverse relation between poverty and real wages. (13)
In the 1980s, when poverty continued to decline but less fast than it did in the 1970s, we have a mixed picture of real wages. In this situation we need to make judgements on the weight of activities in the economy for which the real wage is a proxy. In our view it is the casual wages in agriculture, especially the seasonal ones and the urban casual unskilled wage rates which would affect most of those offering themselves on the labour market. General agricultural/rural wage rates (which reflect the rural non- farm sector as well) showed an increase until the late 1980s; agricultural (seasonal) wages increased on trend in this period. The wages of the unskilled urban workers declined and increased again. There was an increase in skilled workers wages and that of the organized industrial sector. On the whole those real wages which showed a cyclical trend balanced out, while others increased until the late 1980s. The net effect was an increase in real wages and reduction in poverty. This period was one of high growth which was dominantly labour-productivity driven. The inverse relation between real wages and poverty is therefore maintainable for the 1980s.
The 1990s show a decline in real wages of the agricultural seasonally specific wages and a fluctuating cyclical trend in general agricultural/rural wages. Unskilled urban casual wages have also declined. Poverty as we know has probably increased and has certainly not declined in the 1990s. This too confirms the inverse relation between poverty and real wages. It is, however, also evident that the 1990s (like the 1970s) are not a high growth period such as the 1980s were. Growth in the 1990s is less driven by employment expansion than was the case in the 1970s.
In general it may be arguable that in Pakistan, although poverty and real wages are related inversely, as should be the case in an economy in which hired labour constitutes more than half of the employed, both high and low GDP growth situations have led to the alleviation of poverty. However, in the low growth scenario the composition of growth needs to be dominated by employment expansion (as opposed to labour productivity), in order to reduce poverty and improve employment conditions. This is probably not happening in Pakistan in the 1990s. We can now return to examining the employment indicators implicit in the elasticity measure which, as we have seen, had increased from the 1980s to 1990s.
Quite apart from problems of estimation, (14) it is the case that employment elasticities implicitly contain information that may be useful in assessing the employment situation. The overall association of productivity growth and employment growth in any given period is implicit in the elasticity measure by definition. It is, however,m partly determined by technological relationships that obtain within sectors, which are likely to differ across sectors and which are smoothed out or concealed when represented (implicitly) in an elasticity measure or aggregated at an economy-wide level.
It can be argued that during the development process a negative relationship between labour productivity growth and employment growth is to be expected, even a strengthening one. This can be consistent with both employment and productivity increasing at positive rates. Four interesting cases may be defined.
Case 1: When employment and productivity growth are both sizably positive, and there is the likelihood of continuing conditions of healthy economic growth which are not employment unfriendly.
Case 2: When very low (or negative) growth rates in both employment and productivity occur, it may be a sign of a potential sectoral slowdown towards stabilization. This is acceptable, as long as the sector is unlikely to become a natural refuge sector for labour, in times of overall deceleration of growth.
Case 3: When there is negative or very low productivity growth and there is sizeable positive growth in employment, it may be indicative of stagnation in the sector. This is likely to be accompanied by increasing surplus labour and underemployment and falling real wage rates. It is important to note that this may be owing not to an unsatisfactory rate of output growth in the sector but to the employment expansion taking place within it. This is characteristic of a classic refuge sector.
Case 4: When there is positive growth in productivity and a negative or very low growth in employment, it may be indicative of a fundamental technological shift taking place in the sector. (15)
We now discuss productivity and employment growth simultaneously, for the major sectors (Tables A2.5 and A2.6). The following observations can be made from examining the data.
If the indicative schema and observations made above are to be accepted, then the agricultural sector is likely to be moving towards low productivity conditions (although from the productivity perspective it is not as bad as the 1970s) while manufacturing is likely to be undergoing a fundamental technological shift, in which its growth process is becoming de-linked, at least from direct employment generation. Construction is likely to have become a sponge or refuge sector, which is simply absorbing labour at negative productivity levels. There is an increase in employment growth in the retail and trade sector and the 'other' sectors group in the 1990s which is symptomatic of informal and small-scale enterprise sectors growth. This expansion seems associated with relatively lower levels of GDP growth (1970s and 1990s). It has also played an active role in labour absorption in the 1990s, as we have seen earlier.
It is fairly obvious that the growth of the manufacturing sector has strong growth multiplier effects on the rest of the economy (Table A2.7iii), but in looking at employment growth rates for different sectors of the economy we find an interesting set of relationships between employment growth in manufacturing and employment growth in the rest of the economy. Examining simple correlations between sectoral employment growth rates (Table A2.7i), it is clear that growth in manufacturing employment is most strongly related to economy-wide employment growth and, in particular, has positive and relatively stronger linkage effects on employment growth in most other (non-agricultural) sectors of the economy. This is an encouraging finding from the point of view of setting up guidelines for an employment strategy.
The greater the extent to which growth in manufacturing is employment-based, the greater will be its employment multiplier effects on the rest of the economy. However, promoting employment friendly growth does have a potential problem that employment elasticities in manufacturing are low and productivity growth during the 1990s has been associated with a negative employment growth. Hence, the situation in manufacturing is that output growth cannot be expected to generate much direct employment. This is an area which requires special focus in a sector-specific analysis, with special attention paid to the distinctions between the large and small-scale manufacturing sectors.
The trends in the labour force and employment in Pakistan suggest that with the already low and declining participation rates of labour in the economy, labour force growth is itself declining. In this situation, unemployment persists and underemployment is increasing. The indicators of employment conditions, i.e. real wages and poverty, reflect this worsening in the 1990s. In our view, there are also good reasons to expect an increase in labour force growth in the near future. This swelling in the labour force growth is also likely to change the composition of the labour force to a more literate one. In short, the current employment problems facing the economy are likely to magnify in the near future. The major challenge is to attack the problems of unemployment and underemployment and to increase aggregate employment significantly as well.
Apart from controlling population growth, which is an important part of an employment strategy, we need to look closely at the changing features of employment as a whole and in particular sectors of the economy. These issues stand out:
Agriculture in Pakistan has had reasonable output growth in the past. (1) At the aggregate level, its growth was showing signs of decline from the 1980s to the 1990s. (2) On the other hand, employment growth in the sector continued to be positive though it also declined on trend in the 1990s. The sector could be moving towards a situation where it may be forced to absorb large sections of the labour force, possibly unemployable in other sectors that face a slow-down in output growth. This possibility is also reflected in labour absorption in the sector, which showed that it was the main sector 'adjusting' to employment fluctuations in the economy. Part of the reason for this is external to the sector. When the overall economy faces a slowdown, its largest employer (even if its sectoral employment share is declining over the longer run) is likely to experience a swelling in its labour force relative to the output it produces. In the present chapter, we explore these external causes of change as well as what may have happened to the structure of output and employment within the agricultural sector.
Some points need to be noted about employment and output that are specific to agriculture and necessary for interpreting the trends in Pakistan.
Measurement of employment
Employment in agriculture is family-based and also uses hired labour both on a casual/part-time and well as a permanent basis. In other words, employment in agriculture is non-homogenous (3) and does not directly depend on the labour market for all its supply of labour. Seasonal demand for agricultural labour means that those employed, can be expected to be "underemployed" at least some of the time. The notion of full-time work; the role of the labour market in the supply of labour and the aggregation of different types of workers into the employed category are less straightforward in the agricultural sector.
The special nature of output
The determinants of agriculture output are also different from those of output in other sectors. In agriculture, output is partly dependent on external uncertainty, this means that weather is an important determinant of output, especially at low levels of output. Therefore, whenever there are conditions of an immanent collapse of output, labour which has already been employed, appears less productive (productivity declines). Agricultural output collapses are less often failures of demand and are more supply-side determined. Clearly then the measure of labour productivity in such a situation is not so much a proxy of the contribution of labour assuming employment is counted properly) to production but reflects the over-determining influence of other external causes that influence output.
Thus, we need to be cautious in using 'employment' and 'output' categories in themselves or in standard measurements, when analyzing the agricultural sector. We now select two features of this discussion, external uncertainty with output and casual labour use in employment, to posit a combined effect which may be of special relevance in Pakistan.
A combined effect
The degree of external risk in agriculture is also sensitive to the technological features of the production process. The HYV technical input package, in use in Pakistan, which is based on robust varieties of seeds, also requires the use of complementary inputs, (4) the application of which is subject to serious time constraints. (5) It can be argued that the widespread use of the package, though growth- enhancing on average could also increase the sensitivity of the production process to this external uncertainty. If this is the case then it should be reflected not only in an increasing level of output on average but also in relative output instability post-adoption.
One response to increased external risk (due to the profitable technological package and uncertainty), is that land-owning employers move towards flexible arrangements with labour, to reduce a part of their wage-cost risk due to the vagaries of weather. The use of casual labour, (6) particularly during labour demand peaks, can generate this flexibility.
It also means that the greater the proportion of casual labour in the agricultural labour force, the more closely the employment growth trend in agriculture is likely to follow the output growth trend, because it is precisely the casual labour component of the labour force that becomes its variable part.
This would also imply that in conditions of economy-wide low growth, levels of surplus labour in the agrarian economy will be high and underemployment and poverty due to the changes in the composition of the labour force and its casualisation will become increasingly visible.
The growth of output in agriculture has been positive and increasing (tables A3.1 and A3.2, A3.3) in Pakistan. The sector has by and large under-performed GDP growth but its growth rate is reasonable. Figure 3.1 shows that the pattern of output and employment growth has been fluctuating much more since the mid-1980s. Although it is quite true that output growth was much more stable in the 1970s, which meant stable employment growth and stable elasticities, it is likely that the fluctuations apparent from the mid-1980s have contributing factors other than external uncertainty.
These patterns may have something to do with the changing nature of both output and employment. For output, the country-wide use of the HYV technological package by the 1980s (tables A3.3a and 3.3b) meant that its sensitivity to the timing of inputs, as well as external weather conditions with respect to that timing, became more accentuated. This may explain the increasing amplitude of movements in output growth over the years as shown in Figure 3.1. (7)
Because family labour and permanent labour are a constant part of the employed labour force in agriculture, the changes that may result from output booms and slumps are likely to be due to changes in the employment of the casual labour force. We therefore need to examine changes in the composition of the labour force (table A3.4). According to Agricultural Census data agricultural "employment" witnessed a 0.62 per cent annual growth in the 1980s. This excludes casual labour use, which is known to have increased phenomenally in the same period (see table A3.5) (8). Permanent hired labour use has altered but the weight of permanent hired labour in the census figures of permanent agricultural employment is not large enough to warrant comment (2.7 per cent).
Throughout the 1970s the rate of growth of economy-wide employment has been, on a trend, higher than that of agricultural employment. The LFS figures are, as suggested earlier, likely to capture employment in agriculture data, which includes casual labour (table A3.5) and are therefore preferable, at least in terms of coverage, to the Agricultural Census (AC) figures. Both figures for the 1980s (AC and LFS) are, however, lower than the overall growth rate of employment. Since the mid-1980s (which is also the peak period of growth after which a gradual decline sets in), there is no longer a discernible pattern; agricultural employment growth has been higher as well as lower from the overall rate of employment growth (table A3.2). There is also a changing pattern of rural-urban migration, with rural Pakistan increasingly retaining more of the labour force (table A3.6). (9) It is worth noting that for the second half of the 1980s decade, (10) there is a greater increase in growth of 'agricultural employment' than is the case in the first half (table A3.4). (11) Consequently, there are signs of a swelling in the size of the agricultural labour force.
In the 1970s output growth was not too high (2.0 per cent) but it was positive and so was employment (2.1 per cent). This trend altered in the 1980s, when output growth on average was higher (3.9 per cent), and employment growth remained a little lower (1.9 per cent) but switched to an increased use of casual labour. In the 1990s output growth continues at a slightly lower rate (3.3 per cent) and employment growth is also low (1.6 per cent) but both output and employment growth patterns are more volatile (table A3.2).
We now come to the decomposition of growth exercise, which looks at the contribution of labour productivity and employment expansion to output growth, bearing in mind the qualifications made earlier regarding output, employment and productivity in agriculture. Basically, employment expansion-based growth dominated in the 1970s, shifting to productivity-dominated growth in the 1980s, and shifting back to employment expansion based growth in the 1990s, though it is now 'shared' with labour productivity.
In the 1970s, a more stable period of lower output growth, the employment expansion effect dominated the productivity effect for several reasons . The nineteen seventies were a period when the green revolution, although significant, had not spread nationwide (table A3.7a). The popular belief that effective land reform may come about in Pakistan encouraged landed agriculturists with significant holdings to stay in sector. This was also reflected in a spate of land resumptions, a decline in absenteeism and the expansion of cultivable land cultivation. Also, growth in other sectors is this period was low. A combination of these conditions may explain the dominance of the employment effect. (12)
Table 1: Decomposing output growth , productivity and employment effects-agriculture
| Year | Productivity Effect | Employment Effect | Multiple Effect |
| 1969/70 - 1980/81 | 0.063 | 0.918 | 0.018 |
| 1980/81 - 1991/92 | 0.664 | 0.236 | 0.100 |
| 1991/92 - 1994/95 | 0.361 | 0.623 | 0.040 |
The 1980s story is slightly different. Labour productivity dominated contributions to output growth (which was also higher than in the 1970s). Reasons include the less costly expansion of land frontier (already done in the 1970s); the widespread use of a HYV input package (Table A3.7 a and A3.7 b), and greater tractorisation in parts of the country other than the Punjab, where the green revolution had become widespread earlier. The possibility of land reform had also receded from the agenda of the then military regime. But equally significantly, the decline in the employment contribution to growth in the 1980s may be due to higher growth in other sectors of the economy. Sectors that were drawing productive labour out of agriculture particularly manufacturing. It is worth noting that SSE manufacturing, which has a large rural component, was witnessing employment and output growth at least up to the mid-1980s.
The 1990s reverted to an employment expansion based effect (at levels of agricultural output growth similar to the 1980s but more volatile on an annual basis) and may reflect a deceleration of growth in other sectors, particularly manufacturing. There is nonetheless a significant contribution of labour productivity in this period as well.
Employment expansion in the presence of as large a casual labour force as exists in Pakistan in the 1990s says little about the quantum of labour used or its returns. It is worth recalling that the composition of the agricultural labour force has changed in the 1990s compared with the 1980s, and volatility of output in the sector is reflected more in employment growth than previously it is also likely to have generated a precariousness in the livelihoods of those now employed as 'casual labour' in the agricultural sector.
Agricultural employment is a dominant part of rural employment. Trends of unemployment and underemployment in the rural areas reflect labour market conditions in the agricultural sector. However, unemployment, in a developing country context, in particular in agriculture, is not a good indicator of labour market conditions. In the absence of social welfare and insurance, the unemployed are generally available for work in any sector, and often do work to survive. Underemployment is therefore a better indicator of conditions in the labour market. The trends in unemployment and underemployment are examined below.
The data (table A3.8) show that unemployment increased on trend in the 1980s, from 1979-80 to 1987-88. It is not legitimate to observe trends on unemployment without qualification across the year 1990-91 when LFS data is used, as in this year definitional changes took place which particularly affected the unemployed. The data show that unemployment was increasing on trend in rural Pakistan during the 1980s and was showing a trend decline in the 1990s. (13)
The case for rural underemployment is also mixed. It shows a decline from early to mid-1980s and an increasing trend through the second half of the 1980s and the 1990s. (14) What is clear is that the rural underemployment rate is high in Pakistan.
By 1993-94 about 20 per cent of the rural labour force was either not working or only partially working in rural areas, consistent with a fluctuating pattern of growth in output and employment, which as we argued is increasingly based on the use of casual labour. (15) An agricultural sector with low levels of productivity, coexisting with other sectors with limited employment absorption capacities, is likely to have a large reserve army of the underemployed with a high degree of underemployment.
If non-farm employment growth (which is the rural part of the SSE sector) had started slowing down since mid-1980s, then it can be assumed that the bulk of labour absorption in the 1990s would be in the agricultural sector, possibly as underemployment and casual labour. (16)
It has been suggested above that, owing to the non-homogenous nature of the employed, increasing levels of employment (numbers of employed) may not translate into increasing use of labour (amount of labour). Consequently we need to look at the requirements of labour and the many factors which affect them. In the short-run they are best reflected in estimations based on cropping patterns.
Cropping patterns and profitability
Driven by profitability considerations, there has been an increase in the cultivation of crops with a higher labour input in Pakistan. This is clearly a trend across the 1980s and 1990s (table A3.9 ). These changes in cropping patterns which require more labour are also consistent with trends in rates of profits of individual crops (table A3.10 ).
There are two ways to respond to increases in labour requirements. One is by actually increasing labour inputs to meet those requirements. The other is by capital-labour substitutions which may reduce those requirements. Increasing labour inputs can be done by hiring additional labour or by increasing the use of existing labour (either family or hired or both).
Size and productivity: Increasing family labour use
Changes in average sizes of farms may have automatically led to the internalization of some of these increased labour requirements due to changing cropping patterns. The changes in farm size over the intercensal period have increased the incidence of small farms in Pakistan. It can be argued from research on the farm size-productivity relationship in Pakistan, that an inverse relation between farm size and labour input exists (table A3.12 ). (17) So the fall in farm size due to the sub-division of holdings (18) over time would increase the potential availability of labour inputs, especially family labour inputs.
It is also the case that small farms have higher total productivities and greater cropping intensities than larger ones, whereas they may have lower individual-crop yields than large farms ( Mahmood, 1997b, mimeo).
This means that small farms maximize returns through their choice of cropping patterns and greater use of labour. This response to the increasing requirements of labour is one that automatically comes about when productivity led cropping pattern changes made possible by the new technology take place over time. Large farms still have to contend with the hiring-in of extra labour, much of which is likely to be of a casual variety.
Mechanization: Reducing labour requirements
The second way to respond to increasing labour requirements is mechanization (19). Some economy- wide evidence exists here as well. Farm area exclusively using bullock power has declined over the 1980s. On the other hand, farm area using a mix of bullocks and tractors as well as tractors exclusively has increased. It is clear that the effect of tractor use is the displacement of labour (table A3.13).
Tenurial shifts: Creating an available labour pool
Changes in land tenure which have taken place in Pakistan in the last two decades (a decline in tenancy) may or may not have increased productivity. Evidence is mixed and depends on the existence of disincentives in tenancy compared with owner-farming as well as on who is resuming the tenanted land. It is likely that over time disincentives in tenancy have become marginal and insignificant, at least in the Sindh province in Pakistan where the bulk of tenancy exists (Majid, 1994). Reduction in tenancy releases labour into the labour market as landless labour, which needs to get absorbed. Tenancy has also clearly declined in Pakistan (table A3.14).
Results from a recent study (Gazdar, Howe and Zaidi, World Bank, 1994) shows that in rural areas, households headed by tenants and agricultural labourers have a very high incidence of poverty (table A3.15). (20) These workers are at the bottom end of the range of human capital and physical assets categories. This suggests that casual workers constitute one of the poorest sections of the agricultural labour force and, given that tenancy has declined and casual labour increased, these workers are likely to be a growing section of the rural poor.
We have already discussed wage and poverty trends in Chapter Two, but a restatement is in order in the agricultural context. Nominal wage data are available on seasonal agricultural casual workers (table A3.16). Although real wages of agricultural workers improved up to the mid- 1980s, during the 1990s they have probably declined. The wage data again corroborates our expectations about the continuing slow-down of the rural non-farm employment growth since the late-1980s and the swelling of the casual labour force in agriculture. The assumption here is on the nature of the labour market and its extent of isolation. In our view the agricultural casual labour market does function in rural Pakistan; in an overall sense the wage rates are a signal of supply and demand conditions and there is some evidence to support this view. (21) Recent poverty studies in Pakistan also tend to show a decline in rural poverty from 1969-70 up to 1987-88. There is an increase in rural poverty after 1987-88 up to 1990-91, according to one study (Malik, 1996) and a slight decline according to another (Gazdar et. al., 1994), owing to differences in assumptions made to calculate poverty lines (Kemal and Amjad, 1997). However, it can be claimed with some degree of certainty that the declining trend in rural poverty either reversed or witnessed a severe slowdown in 1987-88 to 1990-91, the latest data period available. In our view the increase in rural poverty seems to be a more plausible result.
Agricultural output growth rates were reasonable in the 1980s and 1990s. Employment growth and productivity growth have declined commensurately, leaving a more or less constant elasticity of employment (.48). However, the low employment-low productivity growth situation is sensitive to growth conditions in the rest of the economy and the sector needs to be protected on this count. An overview of the employment situation in agriculture suggests that over the 1970s, 1980s and 1990s, profitability-driven cropping pattern changes based on the new technological package may have increased labour requirements in the sector. The changing distribution of operated holdings, mechanization and tenurial shifts may have partially adapted to, as well as reduced, these increases. The rest of the adjustment has had to be borne by the new hired casual labour force, the size of which is increasing. The nature of this labour force is reflected in the extent of rural underemployment, wage rates and poverty. While underemployment is sizable (around 20 per cent including the unemployed) it has shown signs of an increasing trend since the late 1980s. The returns to casual labour suggest that both the levels of real wages and the size of the non-poor in the rural economy of Pakistan have either stagnated or declined in the last period for which information exists (1987-88 to 1990-91 for poverty and up to 1994-95 for wages).
These changes have been accompanied by what can be seen as structural change in the agricultural sector. This change is manifest both at the level of the production and labour process on the one hand and the supply of agricultural labour on the other. Both these changes are noteworthy from a policy point of view.
First, on the production and the labour process. What is significant about the growth of output and employment in the agricultural sector in Pakistan, is not so much their rates of growth, but the changing composition of the labour force on the one hand and the increased variability in the pattern of its production over time, on the other. There is much greater volatility in agricultural output since the mid-1980s, connected to the widespread assimilation of technical changes in the rural economy and now embedded in the production process. This translates into a fluctuating trend in the employed labour force, with the increasing casual labour component on the receiving end of the adjustments to this volatility in output.
The second matter concerns the expanding and impoverished pool of labour reserve from which this casual labour springs. Landlessness and near-landlessness as a consequence of declining tenancy and the subdivision of holdings is one element in this. The other may be a return labour flow into the agricultural sector because of low growth in other sectors of the economy, especially the rural SSE sector. These tendencies are likely to exacerbate the size of the labour reserve in rural areas, which then has to 'forcibly' get 'absorbed' in agriculture as a last resort.
An effective policy focus to promote future growth and employment in the agricultural sector needs to have a dual aim: the dampening of the relationship between the volatility of output and the livelihoods of the workforce; and making the livelihoods of casual labour within this workforce more sustainable.
In Chapter Two, we concluded that at an aggregate level an employment strategy in Pakistan has to have the manufacturing sector as a critical focus. The argument was based on two observations. First, that given the problems with the elasticity measure, at least a simultaneous examination of employment and labour productivity growth was needed. Second, that cross-sectoral employment linkages also seemed to be important. It was consequently argued that despite the fact that there was limited room within the manufacturing sector to generate employment directly, if manufacturing growth could be employment-based, it would have positive employment effects on the rest of the economy.
Within manufacturing, a distinction can be made between "small-" vs. "large-scale" manufacturing. (1) This section looks at manufacturing as a whole and then focuses on large-scale sector. In a separate chapter (Chapter Five) devoted to the Small-Scale Enterprise Sector (SSE), we look at small-scale manufacturing.
Figure 4.1 gives a visual plot of employment elasticity, as well as employment and output growth rates, in the manufacturing sector over the years in Pakistan. Plotting growth rates of employment and output is useful for an overview and we can start with the same point we made at the economy-wide level about the limitations of elasticity measures taken on their own. A stable and very low elasticity has accompanied fairly diverse output and employment growth patterns in manufacturing over the years in Pakistan. What can be claimed with some degree of confidence is that elasticity of employment is low (2) in manufacturing, and this at least seems to be a structural feature of the sector. The latest period of the 1990s in Pakistan is showing a decline in both output and employment growth and this is a seriously worrying matter for the sector (3). The reason for this low elasticity of employment in Pakistan's manufacturing sector is therefore a critical policy question.
Having observed the trend of output and employment growth, we will now conduct a 'decomposition of growth' exercise for the manufacturing sector. (4) We will first examine how the contributions of labour productivity and employment relate to output growth, and second, whether any trend can be observed in these contributions.
In the first period of the 1970s, although growth itself was uneven (higher in the first half, lower in the second) it was roughly equally divided between productivity and employment. The next period of 1980-81 to 1991-92, the period of consistent high output growth, was based dominantly on the productivity effect (about 81 per cent) while the contribution of employment to growth was low (8 per cent). Clearly, a shift was already taking place in this period of high growth of the manufacturing economy. The last period of the 1990s is one of low and declining growth, especially in the recent years, and the contribution of employment to output growth has been negative. The entire growth process now seems to be based on labour productivity. (5) This trend makes the task of boosting and restructuring the industrial sector, from an employment perspective, somewhat difficult. We need to bear in mind the cross-sectoral employment linkage caveat made in the Chapter Two, to emphasize the role of manufacturing.
Table 4.1: Decomposing output growth: productivity and employment effects - manufacturing
| Year | Productivity Effect | Employment Effect | Multiple Effect |
| 1969/70 - 1980/81 | 0.460 | 0.419 | 0.120 |
| 1980/81 - 1991/92 | 0.816 | 0.089 | 0.094 |
| 1991/92 - 1994/95 | 1.868 | -0.680 | -0.188 |
A low elasticity of employment has been observed in the manufacturing sector in Pakistan. Progressively over the last two decades growth in manufacturing has become more labour-productivity driven (than employment-expansion driven) and in the 1990s it seems to have been de-linked from employment expansion altogether. This picture is quite consistent with the low and declining elasticities of employment already observed. The sources of this growth need to be further explored.
While the share of manufacturing in GDP has increased from 15.1 per cent to 18.2 per cent, its share in employment declined from 13.54 per cent to 10.12 per cent. The rate of growth of value added in the manufacturing sector as a whole has been declining on a trend in Pakistan, over the Plan periods since 1982-83. This is true for manufacturing as a whole as well for large-scale manufacturing. In particular, large-scale manufacturing has experienced a serious slowdown from 1994 onwards and growth was negative in 1996-97. The 1980s were the high growth period of the manufacturing sector in Pakistan. There has been a clear decline, on trend, in the growth of output ever since (6).
Table 4.2: Growth rates of manufacturing output over the sixth, seventh and eighth plan periods
| Period | Total | Large-scale |
| 1982-83 to 1987-88 | 8.2 | 8.1 |
| 1987-88 to 1992-93 | 5.9 | 4.9 |
| 1992-93 to 1996-97 | 3.8 | 1.7 |
| 1992-93 | 5.4 | 4.1 |
| 1993-94 | 5.5 | 4.3 |
| 1994-95 | 3.6 | 1.5 |
| 1995-96 | 4.4 | 2.2 |
| 1996-97 | 1.8 | -1.4 |
Source: Based on Economic Survey (1996-97 and previous issues).
Manufacturing comprises both the small-scale and large-scale sectors in Pakistan. The small-scale sector dominates employment in it, while the large-scale sector dominates output. Table 4.3 shows that although output growth has taken place in the manufacturing sector in both its small and large-scale sections, employment began to decline in the small-scale sector in the mid-1980s. There seems to be some constraint that the small-scale sector begins to face at this time. In other words, the de-linking of employment from growth which was taking place in the manufacturing sector in the 1980s, may have been due to the overwhelming weight of the declining employment in the small-scale manufacturing sector and not because of the large-scale sector as such. The picture after 1990-91 is constructed on the basis of National Income Accounts (NIA) and the labour force surveys (LFS), and there is not much that can be said about the large-scale and small-scale manufacturing sector separately here, except that overall growth of the sector has fallen further, and there has been a negative employment growth. It would seem that at least the same trend with respect to the small-scale sector employment can be assumed.
Table 4.3: Output and employment in manufacturing
| Period | Output | Employment
(Mill. nos.) |
Share of manufacturing sector in total employment | Employment in | Growth rate of total employ-ment | |||
| Total | Large
Scale |
Small
Scale |
Large-scale
manufac. |
Small-scale
manufac. |
||||
| 1982/83 | 45.60 | 33.80 | 11.70 | 3.50 | 13.54 | 0.465 | 3.04 | 0.30 |
| 1983/84 | 49.20 | 36.50 | 12.70 | 3.61 | 13.69 | 0.476 | 3.13 | 3.10 |
| 1984/85 | 53.20 | 39.40 | 13.80 | 3.73 | 13.84 | 0.493 | 3.24 | 3.30 |
| 1985/86 | 57.20 | 42.20 | 15.00 | 3.62 | 13.40 | 0.507 | 3.11 | 2.30 |
| 1986/87 | 61.50 | 45.30 | 16.20 | 4.08 | 14.23 | 0.532 | 3.55 | 12.70 |
| 1987/88 | 67.60 | 50.00 | 17.60 | 3.72 | 12.84 | 0.515 | 3.21 | -8.20 |
| 1988/89 | 70.30 | 51.20 | 19.10 | 3.84 | 12.84 | - | - | 3.20 |
| 1989/90 | 74.30 | 53.70 | 20.70 | 3.96 | 12.84 | - | - | 3.10 |
| 1990/91 | 79.00 | 56.60 | 22.40 | 3.70 | 12.38 | 0.623 | 3.08 | -6.60 |
| 1991/92 | 85.30 | 61.00 | 24.30 | 3.88 | 12.53 | - | - | 4.90 |
| 1992/93 | 89.90 | 63.60 | 26.30 | 3.53 | 11.00 | - | - | -10.00 |
| 1993/94 | 94.70 | 66.20 | 28.50 | 3.34 | 10.12 | - | - | -4.20 |
| 1994/95 | ||||||||