Demographic factors have reappeared in the economic development debate with the emergence of the concept of the "demographic dividend". With many developing countries experiencing a rapid decline in fertility, there has been overwhelming optimism that a demographic bonus will take these countries to greater economic heights. At the same time, there are pessimists doubting the ability of these countries to take advantage of the demographic dividend. This paper looks at the concept critically in the context of India. It tries to empirically estimate the contribution of the age structure change to economic growth in the country through a two-stage least square method. The empirical analysis clearly exhibits a powerful positive impact of the boom in the working age group population on economic growth. This is despite the fact that the educational achievements and health conditions of the people are far from desirable and employment creation is well below the required level.
SPECIAL ARTICLEEconomic & Political Weekly EPW june 21, 200863Glorifying Malthus: Current Debate on ‘Demographic Dividend’ in IndiaK S JamesDemographic factors have reappeared in the economic development debate with the emergence of the concept of the “demographic dividend”.With many developing countries experiencing a rapid decline in fertility, there has been overwhelming optimism that a demographic bonus will take these countries to greater economic heights. At the same time, there are pessimists doubting the ability of these countries to take advantage of the demographic dividend. This paper looks at the concept critically in the context of India. It tries to empirically estimatethe contribution of the age structure change to economic growth in the country through a two-stage least square method. The empirical analysis clearly exhibits a powerful positive impact of the boom in the working age group population on economic growth. This is despite the fact that the educational achievements and health conditions of the people are far from desirable and employment creation is well below the required level. This paper was earlier presented at the Institute for Social and Economic Change, Bangalore and at the annual conference of the Indian Association for the Study of Population held at the Banaras Hindu University, Varanasi during October 26-28, 2007. I have benefited from the comments received in both these occasions. I express my sincere thanks to N Krishnaji, Tiziana Leone, S L Rao, P K Michel Tharakan, S Madeshwaran, B P Vani and an anonymous referee of this journal for critical comments and useful suggestions.K S James (ksjames@gmail.com) is with the population research centre, Institute for Social and Economic Change, Bangalore.Demographic factors have reappeared in the economic development debate with the emergence of the concept of the “demographic dividend”. After the neo-Malthusian onslaught on the detrimental effect of rapid population growth on all aspects of development in the second half of the last cen-tury, the beneficial aspect of population growth has perhaps been sidelined in the development discourse. However, the current changes in the age structure bring population parameters back into the growth debates.The demographic dividend is defined as a rise in the rate of eco-nomic growth due to a rising share of working age people in a population. This phenomenon occurs with a falling birth rate and the consequent shift in the age structure of the population towards the adult working ages. It is also commonly known as the demo-graphic gift or bonus or demographic window. With many devel-oping countries particularly in the Asian continent experiencing a rapid decline in fertility, there has been overwhelming optimism that the demographic bonus will take these countries to greater economic heights [Asian Development Bank 1997; Bloom and Williamson 1998; Cyrus Chu and Lee 2000; Mason 1988]. At the same time there are pessimists doubting the ability of these coun-tries to take advantage of the demographic dividend; they refer to several institutional constraints particularly in providing produc-tive employment to the working age population. It is interesting to note that there has been a tendency in India to valorise the great opportunities existing in the country in terms of demographic dividend by all, irrespective of ideology although the concept itself is deep-rooted in the neo-Malthusian framework [Chandrasekhar et al 2006; Mitra and Nagarajan 2005; Narayana 2007]. The purpose of this paper is to look at the concept of the demo-graphic dividend more critically in the context of India. It also tries to empirically estimate the contribution of the age structure change to economic growth in the country through a two-stage least square method to overcome the well known endogenity issue on the relationship between population change and economic development. Although robust estimates are hard to derive due to want of adequate data on different aspects of the economy particularly at the state level, the estimates provide a useful insight on the demography-economy relationship and place the debate in a proper context. 1 The Concept of Demographic DividendThere are three phases of the demographic transition process creating three unique age structures for any country. In the first phase, with fertility being very high and mortality declining,
SPECIAL ARTICLEjune 21, 2008 EPW Economic & Political Weekly64there will be a large number of people in the young age group, particularly below 15 years, creating a high dependency ratio. In the second phase of transition, the fertility starts declining at a fast pace leading to a reduction in the child population. However, as a consequence of higher fertility in the past, there will continue to be higher growth rate in the working age group population during this period. This period is marked by considerable reduc-tion in the dependency ratio. In the third phase, the dependency ratio again will be higher as a result of higher old age population. The demographic bonus or gift is a term used to understand the second phase of the age structure transition. The proportion of the working age population in the total population will be one of the highest during this period with a potential growth inducing impact as well. This is expected to be a short phase in any popula-tion depending upon the pace of fertility decline. If the fertility decline is slow and steady as occurred in the western countries, this phase may even pass unnoticed. But for the developing coun-tries of today, which are experiencing a rapid and sudden decline in fertility, the age structure transition is vividly evident and is expected to be present for over 40 years.However, it should be remembered that the growth of the working age population may be higher even in the first stage of the age structure transition along with a high growth of child population. Hence, the advantage in the second stage of the age structure transition is mainly relative in nature and not in abso-lute terms. The growth of the working age population will be higher in this period as compared to the growth of child popula-tion, leading to a lower dependency ratio. “The young and old tend to consume more output than they generate, unlike working age individuals, whose contribution to output and saving tends to be more than commensurate with their consumption” [Bloom, Canning and Malaney 2000]. It is often argued that the contribution of the second stage of the age structure transition depends on the ability of countries to take full advantage of this stage in providing gainful employment to the workforce. Also the age structure transition due to a dras-tic decline in fertility provides several economic benefits particu-larly at the household level. Therefore, it is also necessary to look more closely into other pathways by which a reduction in the dependency rate could affect the economy.Demographic Rationale of Economic ChangeThe age structure transition can create a spurt in economic growth in several ways. First and foremost is the increased saving expected during the age structure transition [Mason 1988; Higgins and Williamson 1997; Deaton and Paxson 1997; Lee et al 2000]. The increase in saving rate happens primarily due to the low dependency rate and partly also due to increased life expect-ancy. The major part of this saving is definitely the contribution by households. In a way, the increased saving is not due to a large number of adult members but due to a fewer number of children because the number of adults would have been the same even in the earlier stages of age structure transition but coupled with a higher number of children.Secondly, with the decline in fertility, women are more likely to enter into the labour market during this stage. This will result in increased economic activity and would lead to a spurt in economic growth. One of the major hurdles to women entering the labour market had been the high fertility and the time to be spent on childcare. A decline in fertility gives women the freedom to spend more time on economic activities resulting in higher economic growth during this period [Becker and Lewis 1973].Thirdly, it is also pointed out that people invest more on their own health when children are fewer in number, leading to better productivity and economic benefits to the household. There is now an increasing recognition of the advantages of better health for increasing productivity and thus better economic outcomes. Lucas (1993) emphasised productivity growth as the source of economic miracles in east Asian countries.Yet another argument in favour of demographic dividend is that the government also will be in a position to spend and invest in more productive activities with the decline in the number of children as public spending on education and health can be diverted to more productive activities [World Bank 1984].It is clear from the above that the economic advantages of demographic dividend is primarily achieved at the household level mainly due to fertility change and gets aggregated into a macrolevel advantage. However, the discussion on demographic dividend often does not consider the importance of the fertility transition advantage and limits its discussion to the age structure transition. There seems to be also a tendency to consider the macroeconomic policy environment as the crucial force deter-mining the relevant benefits. Demographic Change and Economic GrowthAlthough the evidence on the relationship between demographic change and economic development is drawn only from a limited number of countries, most of the studies show a strong positive association between demographic variables and economic out-comes. The analysis by Bloom and Williamson (1998) of 78 Asian and non-Asian countries showed a powerful positive impact of growth of the working age population on economic growth. The estimates showed that nearly one-third of the economic miracle of east Asian countries can be attributed to the demo-graphic dividend. Similarly Behrman et al (1999) using panel data for several countries since 1950 looked at the relationship between the aver-age age of the population and several economic outcomes. The study found a strong positive association between the age pattern and economic outcome. Andersson’s (2001) study of Scandinavian countries using data since 1980 also found a positive association between the share of economic growth and the share of the work-ing age population. The case of Ireland, which was lagging in fertility transition in Europe, needs special mention. With ferti-lity starting to decline faster in the 1980s, the country also entered into a phase of augmented economic growth indicating a clear association between demographic change and economic growth. Bloom et al (2003) and Bloom et al (2006) using a panel data of countries from 1960 to 2000, established a positive asso-ciation between the age structure transition and economic growth in India and China. They have also predicted higher growth prospects for India compared to China over the next 30
SPECIAL ARTICLEEconomic & Political Weekly EPW june 21, 200865years, as the effect of the fertility decline and the bulge of popu-lation age cohort in the working age group will be sharper in India in the coming decades.On the other hand, some not-very-optimistic views on the rela-tionship between age structure and economic development are also observed in the case of developing countries. Navaneetham (2002) found a positive impact on economic growth of age struc-ture in south-east Asia but failed to see such a relationship in south Asian countries. Another often reported instance is the Latin American experience where the age structure transition seems to have failed to augment economic growth [Bloom et al 2003]. In the case of India it is argued that that the demographic changes are not sufficient to provide an upward push to the rate of economic growth [Mitra and Nagarajan 2005; Chandrasekhar et al 2006]. Interestingly, even the pessimists on the demographic dividend do not rule out the beneficial effects of age structure change but are mainly worried about the existing institutional bottlenecks in some countries that make it difficult to achieve these benefits. There is a near universal agreement that there is nothing auto-matic about the links between demographic change and economic growth. For instance, Navaneetham (2002) felt that the lack of a clear cut association between demographic change and economic growth in south Asia, compared to south-east Asia, is primarily due to a lack of openness of the former economies to trade. Bloom et al (2006) says that the age distribution changes create supply side potential but whether the potential is realised depends upon the policy environment in each country, particularly the open-ness to trade. According to Chandrasekhar et al (2006), with India facing a major deficit in the area of education and health, the conversion of a growing labour force into a quality workforce is difficult to achieve; this will ultimately result in a waste of the demographic advantage.However, it should be remembered that the pathways by which demographic variables impact on the economy operates at the household level and need not necessarily have a direct connec-tion with the macroeconomic environment. Among the factors contributing to economic growth during the demographic dividend stage, the household saving effect and the health-productivity links are automatic and have to be purely attributed to the fertility transition. Perhaps, the female labour force participation depends also upon the policy environment. However, it is well known that many states in India have not made any headway in providing effective employment to women, and even without that the household economy should benefit from the demographic dividend. Therefore, it is necessary to see empirically how far the demographic dividend benefits India.2 TheTheoreticalConfusionThere seems to be some confusion in placing the demographic dividend argument within the overall debate on population and development, which has a long and rich history ever since Malthusian pessimism took root. It is interesting to note that even when there is a near unanimity in accepting the advantages of a demographic dividend in India, there is a considerable confusion on the theoretical premises of this argument. For instance, Chandrasekhar et al (2006) considered that “the notion of ‘demo-graphic dividend’ overturns the older popular perception that the large and excess population is a problem rather than a benefit from an economic point of view”. The concept of the demographic dividend primarily emerges from the neo-Malthusian analysis of the adverse impact of rapid population growth on economic development. In their classic work Coale and Hoover (1958) brought out in unequivocal terms the detrimental impact of some demographic factors on economic growth. They have particularly mentioned three demographic forces adversely affecting development: the size of the popula-tion, growth rate and age structure. These three forces have three different types of impact on the economy. Firstly, there will be a capital shallowing effect as rapid population growth leads to a fall in the ratio of capital to labour. Secondly, the age-dependency effect creates a worsening dependency ratio due to a rise in the young population which will ultimately erode the savings in the household. Finally, the investment diversion effect leads to a large amount of money being spent by the government on the social sector rather than for productive, growth-oriented invest-ment. The Coale and Hoover (1958) argument suggests that declines in fertility promote growth through decreases in the dependency ratios.However, the neo-Malthusian onslaught on the adverse conse-quence of population increase has been partly silenced by the revisionists in the 1980s, who argued that population growth is not a major deterrent to economic development [Kelly 1988; National Research Council 1986]. They suggested that popula-tion growth has both negative and positive impacts on develop-ment. The net impact depends upon the specific conditions in each country and no universal conclusion can be derived on this relationship. The revisionists, thus, downgraded the relative importance of population growth as an important variable determining economic development [Kelly 2001]. Although there had been wide criticism of the Coale and Hoover argument that projects a pessimistic view of the impact of population growth in India, there seems suddenly a unanimity in the possibilities of a favourable impact of the demographic dividend, although both sets of arguments are based upon the same premises. The only point of disagreement seems to relate to whether India will be able to take advantage of the demographic dividend with low literacy levels and several other adverse human development indicators.3 Empirical Estimation of Demographic DividendThe Indian states with a wide variation in demographic achieve-ments present interesting data that enable us to understand the implications of the demographic dividend. Empirical estimation of the demographic dividend is methodologically challenging due to the lack of adequate data at the state level. As already pointed out the relationship between age structure transition and economic change is mediated by factors like the saving rate, female labour force participation, etc. Therefore, before specifi-cally getting into the econometric estimation, it is necessary to look at the trends of these intervening factors during the various stages of age structure transition.
60+ age group 15 59 age group 0-14 age group
1981 91 1991-2001
SPECIAL ARTICLEjune 21, 2008 EPW Economic & Political Weekly68Table 3: Female Labour Force Participation in Major States(15-59 age group, 1981-2001, in %)State 198119912001Andhra Pradesh 52.21 52.56 51.89Bihar 22.6425.4235.04Gujarat 33.23 41.33 42.56Haryana 17.7818.5643.60Karnataka 40.66 46.19 47.80Kerala 26.6324.0422.45Madhya Pradesh 48.87 52.61 46.93Maharashtra 49.89 53.69 56.93Orissa 31.73 32.80 37.70Punjab 9.777.0128.66Rajasthan 35.5746.0554.99Tamil Nadu 39.78 43.99 44.26Uttar Pradesh 13.39 20.76 28.28West Bengal 13.38 18.18 28.13India 31.85 35.86 40.02Source: Computed from Different Census Data.Table 4: Determinants of Growth Rate of Income Per CapitaSl No Independent Variables Beta Coefficient OLS Beta Coefficient 2SLS (1) (2)1 Constant 5.56 (18.75) -2.83 (14.94)2 Log initial per capita income (1971) -0.97 (2.08) -0.81 (1.72)3 Life expectancy 0.03 (0.09) -0.76* (0.15)4 Factor of adult literacy and non-primary sector employment 1.44* (0.66) 5.23* (0.88)5 Log female labour force participation 1.15 (0.62) 1.62* (0.47)6 Growth rate of working age population -0.35 (0.86) –7 States with poor age structure transition (northern states and Assam) Ref: southern states 1.01 (1.09) 2.26* (0.90)8 States with medium age structure transition (other non-southern states) Ref: southern states 2.25* (1.04) 1.99* (0.83)9 Growth rate of overall population – -2.28* (0.96)10 Instrumental variable for growth rate of working age population – 24.19* (4.19) R-Square (per cent) 36 62 Prob > F 0.0009 0.0000 No of cases 60 60Standard error is provided in parenthesis. * Significant at 5 per cent level.account for the wide differences in the demographic transition, the analysis also included dummy variables for groups of states. The low demographic-indicator states consist of four northern states (Bihar, Madhya Pradesh, Rajasthan and Uttar Pradesh) and Assam and the high demographic indicator states consist of all the southern states. All other states are included in the middle group. One of the important variables that mediate the demography-economic growth relationship is the household saving rate. How-ever, data on saving rate are not available at the state level inIndia. Hence the estimates are made without considering household saving information. The econometric method used for the estimation is the two-stage least square (2SLS) method. The aim of the econometric analysis, as pointed out earlier, is to see how far the age structure variable affects economic growth in the country. We have modelled the growth rate of per capita income with growth rate of working age popu-lation and other important variables, similar to the model adopt-ed for different countries by others [Bloom and Williamson 1998, Bloom et al 2006]. The purpose of using a 2SLS method is to address the endogenity issue often observed between population growth and economic change. It is argued that the failure to consider the possibility of reverse causality between population growth and economic development often leads to erroneous conclusions from the empirical estimates [Bloom and Williamson 1998]. Many studies conducted in recent years to understand the demographic dividend have found that the ordinary least-square (OLS) regres-sion estimates are inadequate to explain the age structure impact [Bloom and Williamson 1998, Bloom et al 2006]. To find out the appropriateness of using panel data approach for estimating parameters by random-effect mod-el, we have carried out the Breusch and Pagan (1980) Lagrangian multiplier test and found lack of systematic variation between ui’s (neither with respect to state nor with year) to justify the appli-cation of random-effect model. Hence the estima-tion is carried out using OLS and 2SLS methods for time series-cross section analysis. The full model adopted for the estimation is Y1 = α1 + Σβjxj + δ1Y2 + e1 (1) jwhere Y2 estimated as Y2 = α2 + Σλjzj + δ2Y1 + e2 (2) jwhere Y1 is the growth rate of per capita income, and xj are the vector independent variables influencing Y1 except the growth of adult population. The adult population growth is entered as an instrumental variable in the model (δ2Y1) derived from the second model. Y2 is the growth rate of working age population and zj’s are the vector of independent variables such as life expectancy and adult literacy rate. The sec-ond model also includes the growth rate of per capita income (Y1). The e1 and e2 are the error term of the respective models.As a first step, we have carried out anOLS estimate without an instrumental variable and this is presented in column 1 in Table 4. Due to the high correlation of the independent variables such as the adult literacy rate, percentage of non-agricultural workers and life expectancy, we have used a factor of adult literacy and non-agricultural workers in the model. The estimates show that most of the variables entered into the model are not significant. The only variable showing significant expected relationship is the factor of adult literacy-non-agricultural workers and the regional dummy for the states with medium age structure transi-tion. The age structure variable also does not show a significant relationship with per capita income growth. This may be perhaps due to reverse causality and we have adopted a 2SLS method for estimation, which is presented in column 2 of Table 4. Most variables included in the second model estimated using 2SLS method show a significant relationship with growth of per capita income. The predictability of the model has also substan-tially improved as compared to the OLS estimates. The adult literacy-non-primary sector workers factor and female labour force participation, as expected, enhance economic growth potential. Contrary to this, life expectancy shows a negative relationship with growth rate of income per capita. Life expectancy has regis-tered a marked improvement in most states in India with drastic reduction in mortality throughout the country. A negative rela-tionship of life expectancy with economic growth, therefore, is not completely unexpected as mortality improvement has taken place irrespective of economic growth in many states. The initial level of per capita income is found to be unimportant deciding the pace of economic progress. The age structure variable has shown a significant positive relationship with economic growth. It indicates that the age structure changes, as in the case of other countries, have a strong potential to enhance economic growth in India as well. We have
SPECIAL ARTICLEEconomic & Political Weekly EPW june 21, 200863Glorifying Malthus: Current Debate on ‘Demographic Dividend’ in IndiaK S JamesDemographic factors have reappeared in the economic development debate with the emergence of the concept of the “demographic dividend”.With many developing countries experiencing a rapid decline in fertility, there has been overwhelming optimism that a demographic bonus will take these countries to greater economic heights. At the same time, there are pessimists doubting the ability of these countries to take advantage of the demographic dividend. This paper looks at the concept critically in the context of India. It tries to empirically estimatethe contribution of the age structure change to economic growth in the country through a two-stage least square method. The empirical analysis clearly exhibits a powerful positive impact of the boom in the working age group population on economic growth. This is despite the fact that the educational achievements and health conditions of the people are far from desirable and employment creation is well below the required level. This paper was earlier presented at the Institute for Social and Economic Change, Bangalore and at the annual conference of the Indian Association for the Study of Population held at the Banaras Hindu University, Varanasi during October 26-28, 2007. I have benefited from the comments received in both these occasions. I express my sincere thanks to N Krishnaji, Tiziana Leone, S L Rao, P K Michel Tharakan, S Madeshwaran, B P Vani and an anonymous referee of this journal for critical comments and useful suggestions.K S James (ksjames@gmail.com) is with the population research centre, Institute for Social and Economic Change, Bangalore.Demographic factors have reappeared in the economic development debate with the emergence of the concept of the “demographic dividend”. After the neo-Malthusian onslaught on the detrimental effect of rapid population growth on all aspects of development in the second half of the last cen-tury, the beneficial aspect of population growth has perhaps been sidelined in the development discourse. However, the current changes in the age structure bring population parameters back into the growth debates.The demographic dividend is defined as a rise in the rate of eco-nomic growth due to a rising share of working age people in a population. This phenomenon occurs with a falling birth rate and the consequent shift in the age structure of the population towards the adult working ages. It is also commonly known as the demo-graphic gift or bonus or demographic window. With many devel-oping countries particularly in the Asian continent experiencing a rapid decline in fertility, there has been overwhelming optimism that the demographic bonus will take these countries to greater economic heights [Asian Development Bank 1997; Bloom and Williamson 1998; Cyrus Chu and Lee 2000; Mason 1988]. At the same time there are pessimists doubting the ability of these coun-tries to take advantage of the demographic dividend; they refer to several institutional constraints particularly in providing produc-tive employment to the working age population. It is interesting to note that there has been a tendency in India to valorise the great opportunities existing in the country in terms of demographic dividend by all, irrespective of ideology although the concept itself is deep-rooted in the neo-Malthusian framework [Chandrasekhar et al 2006; Mitra and Nagarajan 2005; Narayana 2007]. The purpose of this paper is to look at the concept of the demo-graphic dividend more critically in the context of India. It also tries to empirically estimate the contribution of the age structure change to economic growth in the country through a two-stage least square method to overcome the well known endogenity issue on the relationship between population change and economic development. Although robust estimates are hard to derive due to want of adequate data on different aspects of the economy particularly at the state level, the estimates provide a useful insight on the demography-economy relationship and place the debate in a proper context. 1 The Concept of Demographic DividendThere are three phases of the demographic transition process creating three unique age structures for any country. In the first phase, with fertility being very high and mortality declining, 
SPECIAL ARTICLEjune 21, 2008 EPW Economic & Political Weekly64there will be a large number of people in the young age group, particularly below 15 years, creating a high dependency ratio. In the second phase of transition, the fertility starts declining at a fast pace leading to a reduction in the child population. However, as a consequence of higher fertility in the past, there will continue to be higher growth rate in the working age group population during this period. This period is marked by considerable reduc-tion in the dependency ratio. In the third phase, the dependency ratio again will be higher as a result of higher old age population. The demographic bonus or gift is a term used to understand the second phase of the age structure transition. The proportion of the working age population in the total population will be one of the highest during this period with a potential growth inducing impact as well. This is expected to be a short phase in any popula-tion depending upon the pace of fertility decline. If the fertility decline is slow and steady as occurred in the western countries, this phase may even pass unnoticed. But for the developing coun-tries of today, which are experiencing a rapid and sudden decline in fertility, the age structure transition is vividly evident and is expected to be present for over 40 years.However, it should be remembered that the growth of the working age population may be higher even in the first stage of the age structure transition along with a high growth of child population. Hence, the advantage in the second stage of the age structure transition is mainly relative in nature and not in abso-lute terms. The growth of the working age population will be higher in this period as compared to the growth of child popula-tion, leading to a lower dependency ratio. “The young and old tend to consume more output than they generate, unlike working age individuals, whose contribution to output and saving tends to be more than commensurate with their consumption” [Bloom, Canning and Malaney 2000]. It is often argued that the contribution of the second stage of the age structure transition depends on the ability of countries to take full advantage of this stage in providing gainful employment to the workforce. Also the age structure transition due to a dras-tic decline in fertility provides several economic benefits particu-larly at the household level. Therefore, it is also necessary to look more closely into other pathways by which a reduction in the dependency rate could affect the economy.Demographic Rationale of Economic ChangeThe age structure transition can create a spurt in economic growth in several ways. First and foremost is the increased saving expected during the age structure transition [Mason 1988; Higgins and Williamson 1997; Deaton and Paxson 1997; Lee et al 2000]. The increase in saving rate happens primarily due to the low dependency rate and partly also due to increased life expect-ancy. The major part of this saving is definitely the contribution by households. In a way, the increased saving is not due to a large number of adult members but due to a fewer number of children because the number of adults would have been the same even in the earlier stages of age structure transition but coupled with a higher number of children.Secondly, with the decline in fertility, women are more likely to enter into the labour market during this stage. This will result in increased economic activity and would lead to a spurt in economic growth. One of the major hurdles to women entering the labour market had been the high fertility and the time to be spent on childcare. A decline in fertility gives women the freedom to spend more time on economic activities resulting in higher economic growth during this period [Becker and Lewis 1973].Thirdly, it is also pointed out that people invest more on their own health when children are fewer in number, leading to better productivity and economic benefits to the household. There is now an increasing recognition of the advantages of better health for increasing productivity and thus better economic outcomes. Lucas (1993) emphasised productivity growth as the source of economic miracles in east Asian countries.Yet another argument in favour of demographic dividend is that the government also will be in a position to spend and invest in more productive activities with the decline in the number of children as public spending on education and health can be diverted to more productive activities [World Bank 1984].It is clear from the above that the economic advantages of demographic dividend is primarily achieved at the household level mainly due to fertility change and gets aggregated into a macrolevel advantage. However, the discussion on demographic dividend often does not consider the importance of the fertility transition advantage and limits its discussion to the age structure transition. There seems to be also a tendency to consider the macroeconomic policy environment as the crucial force deter-mining the relevant benefits. Demographic Change and Economic GrowthAlthough the evidence on the relationship between demographic change and economic development is drawn only from a limited number of countries, most of the studies show a strong positive association between demographic variables and economic out-comes. The analysis by Bloom and Williamson (1998) of 78 Asian and non-Asian countries showed a powerful positive impact of growth of the working age population on economic growth. The estimates showed that nearly one-third of the economic miracle of east Asian countries can be attributed to the demo-graphic dividend. Similarly Behrman et al (1999) using panel data for several countries since 1950 looked at the relationship between the aver-age age of the population and several economic outcomes. The study found a strong positive association between the age pattern and economic outcome. Andersson’s (2001) study of Scandinavian countries using data since 1980 also found a positive association between the share of economic growth and the share of the work-ing age population. The case of Ireland, which was lagging in fertility transition in Europe, needs special mention. With ferti-lity starting to decline faster in the 1980s, the country also entered into a phase of augmented economic growth indicating a clear association between demographic change and economic growth. Bloom et al (2003) and Bloom et al (2006) using a panel data of countries from 1960 to 2000, established a positive asso-ciation between the age structure transition and economic growth in India and China. They have also predicted higher growth prospects for India compared to China over the next 30 
SPECIAL ARTICLEEconomic & Political Weekly EPW june 21, 200865years, as the effect of the fertility decline and the bulge of popu-lation age cohort in the working age group will be sharper in India in the coming decades.On the other hand, some not-very-optimistic views on the rela-tionship between age structure and economic development are also observed in the case of developing countries. Navaneetham (2002) found a positive impact on economic growth of age struc-ture in south-east Asia but failed to see such a relationship in south Asian countries. Another often reported instance is the Latin American experience where the age structure transition seems to have failed to augment economic growth [Bloom et al 2003]. In the case of India it is argued that that the demographic changes are not sufficient to provide an upward push to the rate of economic growth [Mitra and Nagarajan 2005; Chandrasekhar et al 2006]. Interestingly, even the pessimists on the demographic dividend do not rule out the beneficial effects of age structure change but are mainly worried about the existing institutional bottlenecks in some countries that make it difficult to achieve these benefits. There is a near universal agreement that there is nothing auto-matic about the links between demographic change and economic growth. For instance, Navaneetham (2002) felt that the lack of a clear cut association between demographic change and economic growth in south Asia, compared to south-east Asia, is primarily due to a lack of openness of the former economies to trade. Bloom et al (2006) says that the age distribution changes create supply side potential but whether the potential is realised depends upon the policy environment in each country, particularly the open-ness to trade. According to Chandrasekhar et al (2006), with India facing a major deficit in the area of education and health, the conversion of a growing labour force into a quality workforce is difficult to achieve; this will ultimately result in a waste of the demographic advantage.However, it should be remembered that the pathways by which demographic variables impact on the economy operates at the household level and need not necessarily have a direct connec-tion with the macroeconomic environment. Among the factors contributing to economic growth during the demographic dividend stage, the household saving effect and the health-productivity links are automatic and have to be purely attributed to the fertility transition. Perhaps, the female labour force participation depends also upon the policy environment. However, it is well known that many states in India have not made any headway in providing effective employment to women, and even without that the household economy should benefit from the demographic dividend. Therefore, it is necessary to see empirically how far the demographic dividend benefits India.2 TheTheoreticalConfusionThere seems to be some confusion in placing the demographic dividend argument within the overall debate on population and development, which has a long and rich history ever since Malthusian pessimism took root. It is interesting to note that even when there is a near unanimity in accepting the advantages of a demographic dividend in India, there is a considerable confusion on the theoretical premises of this argument. For instance, Chandrasekhar et al (2006) considered that “the notion of ‘demo-graphic dividend’ overturns the older popular perception that the large and excess population is a problem rather than a benefit from an economic point of view”. The concept of the demographic dividend primarily emerges from the neo-Malthusian analysis of the adverse impact of rapid population growth on economic development. In their classic work Coale and Hoover (1958) brought out in unequivocal terms the detrimental impact of some demographic factors on economic growth. They have particularly mentioned three demographic forces adversely affecting development: the size of the popula-tion, growth rate and age structure. These three forces have three different types of impact on the economy. Firstly, there will be a capital shallowing effect as rapid population growth leads to a fall in the ratio of capital to labour. Secondly, the age-dependency effect creates a worsening dependency ratio due to a rise in the young population which will ultimately erode the savings in the household. Finally, the investment diversion effect leads to a large amount of money being spent by the government on the social sector rather than for productive, growth-oriented invest-ment. The Coale and Hoover (1958) argument suggests that declines in fertility promote growth through decreases in the dependency ratios.However, the neo-Malthusian onslaught on the adverse conse-quence of population increase has been partly silenced by the revisionists in the 1980s, who argued that population growth is not a major deterrent to economic development [Kelly 1988; National Research Council 1986]. They suggested that popula-tion growth has both negative and positive impacts on develop-ment. The net impact depends upon the specific conditions in each country and no universal conclusion can be derived on this relationship. The revisionists, thus, downgraded the relative importance of population growth as an important variable determining economic development [Kelly 2001]. Although there had been wide criticism of the Coale and Hoover argument that projects a pessimistic view of the impact of population growth in India, there seems suddenly a unanimity in the possibilities of a favourable impact of the demographic dividend, although both sets of arguments are based upon the same premises. The only point of disagreement seems to relate to whether India will be able to take advantage of the demographic dividend with low literacy levels and several other adverse human development indicators.3 Empirical Estimation of Demographic DividendThe Indian states with a wide variation in demographic achieve-ments present interesting data that enable us to understand the implications of the demographic dividend. Empirical estimation of the demographic dividend is methodologically challenging due to the lack of adequate data at the state level. As already pointed out the relationship between age structure transition and economic change is mediated by factors like the saving rate, female labour force participation, etc. Therefore, before specifi-cally getting into the econometric estimation, it is necessary to look at the trends of these intervening factors during the various stages of age structure transition. 
60+ age group 15 59 age group 0-14 age group 
1981 91 1991-2001 
SPECIAL ARTICLEjune 21, 2008 EPW Economic & Political Weekly68Table 3: Female Labour Force Participation in Major States(15-59 age group, 1981-2001, in %)State 198119912001Andhra Pradesh 52.21 52.56 51.89Bihar 22.6425.4235.04Gujarat 33.23 41.33 42.56Haryana 17.7818.5643.60Karnataka 40.66 46.19 47.80Kerala 26.6324.0422.45Madhya Pradesh 48.87 52.61 46.93Maharashtra 49.89 53.69 56.93Orissa 31.73 32.80 37.70Punjab 9.777.0128.66Rajasthan 35.5746.0554.99Tamil Nadu 39.78 43.99 44.26Uttar Pradesh 13.39 20.76 28.28West Bengal 13.38 18.18 28.13India 31.85 35.86 40.02Source: Computed from Different Census Data.Table 4: Determinants of Growth Rate of Income Per CapitaSl No Independent Variables Beta Coefficient OLS Beta Coefficient 2SLS (1) (2)1 Constant 5.56 (18.75) -2.83 (14.94)2 Log initial per capita income (1971) -0.97 (2.08) -0.81 (1.72)3 Life expectancy 0.03 (0.09) -0.76* (0.15)4 Factor of adult literacy and non-primary sector employment 1.44* (0.66) 5.23* (0.88)5 Log female labour force participation 1.15 (0.62) 1.62* (0.47)6 Growth rate of working age population -0.35 (0.86) –7 States with poor age structure transition (northern states and Assam) Ref: southern states 1.01 (1.09) 2.26* (0.90)8 States with medium age structure transition (other non-southern states) Ref: southern states 2.25* (1.04) 1.99* (0.83)9 Growth rate of overall population – -2.28* (0.96)10 Instrumental variable for growth rate of working age population – 24.19* (4.19) R-Square (per cent) 36 62 Prob > F 0.0009 0.0000 No of cases 60 60Standard error is provided in parenthesis. * Significant at 5 per cent level.account for the wide differences in the demographic transition, the analysis also included dummy variables for groups of states. The low demographic-indicator states consist of four northern states (Bihar, Madhya Pradesh, Rajasthan and Uttar Pradesh) and Assam and the high demographic indicator states consist of all the southern states. All other states are included in the middle group. One of the important variables that mediate the demography-economic growth relationship is the household saving rate. How-ever, data on saving rate are not available at the state level inIndia. Hence the estimates are made without considering household saving information. The econometric method used for the estimation is the two-stage least square (2SLS) method. The aim of the econometric analysis, as pointed out earlier, is to see how far the age structure variable affects economic growth in the country. We have modelled the growth rate of per capita income with growth rate of working age popu-lation and other important variables, similar to the model adopt-ed for different countries by others [Bloom and Williamson 1998, Bloom et al 2006]. The purpose of using a 2SLS method is to address the endogenity issue often observed between population growth and economic change. It is argued that the failure to consider the possibility of reverse causality between population growth and economic development often leads to erroneous conclusions from the empirical estimates [Bloom and Williamson 1998]. Many studies conducted in recent years to understand the demographic dividend have found that the ordinary least-square (OLS) regres-sion estimates are inadequate to explain the age structure impact [Bloom and Williamson 1998, Bloom et al 2006]. To find out the appropriateness of using panel data approach for estimating parameters by random-effect mod-el, we have carried out the Breusch and Pagan (1980) Lagrangian multiplier test and found lack of systematic variation between ui’s (neither with respect to state nor with year) to justify the appli-cation of random-effect model. Hence the estima-tion is carried out using OLS and 2SLS methods for time series-cross section analysis. The full model adopted for the estimation is Y1 = α1 + Σβjxj + δ1Y2 + e1 (1) jwhere Y2 estimated as Y2 = α2 + Σλjzj + δ2Y1 + e2 (2) jwhere Y1 is the growth rate of per capita income, and xj are the vector independent variables influencing Y1 except the growth of adult population. The adult population growth is entered as an instrumental variable in the model (δ2Y1) derived from the second model. Y2 is the growth rate of working age population and zj’s are the vector of independent variables such as life expectancy and adult literacy rate. The sec-ond model also includes the growth rate of per capita income (Y1). The e1 and e2 are the error term of the respective models.As a first step, we have carried out anOLS estimate without an instrumental variable and this is presented in column 1 in Table 4. Due to the high correlation of the independent variables such as the adult literacy rate, percentage of non-agricultural workers and life expectancy, we have used a factor of adult literacy and non-agricultural workers in the model. The estimates show that most of the variables entered into the model are not significant. The only variable showing significant expected relationship is the factor of adult literacy-non-agricultural workers and the regional dummy for the states with medium age structure transi-tion. The age structure variable also does not show a significant relationship with per capita income growth. This may be perhaps due to reverse causality and we have adopted a 2SLS method for estimation, which is presented in column 2 of Table 4. Most variables included in the second model estimated using 2SLS method show a significant relationship with growth of per capita income. The predictability of the model has also substan-tially improved as compared to the OLS estimates. The adult literacy-non-primary sector workers factor and female labour force participation, as expected, enhance economic growth potential. Contrary to this, life expectancy shows a negative relationship with growth rate of income per capita. Life expectancy has regis-tered a marked improvement in most states in India with drastic reduction in mortality throughout the country. A negative rela-tionship of life expectancy with economic growth, therefore, is not completely unexpected as mortality improvement has taken place irrespective of economic growth in many states. The initial level of per capita income is found to be unimportant deciding the pace of economic progress. The age structure variable has shown a significant positive relationship with economic growth. It indicates that the age structure changes, as in the case of other countries, have a strong potential to enhance economic growth in India as well. We have 
