aa r X i v : . [ ec on . GN ] J a n Social Mobility in India
A. Singh, Department of Economics and Finance, Pilani Campus, India,email: [email protected]. Forcina, Dipartimento di Economia, Perugia, ItalyK. Muniyoor, Department of Economics and Finance, Pilani Campus, IndiaFebruary 2, 2021
Abstract
Rapid rise in income inequality in India is a serious concern. While the emphasis is on inclusivegrowth, it seems difficult to tackle the problem without looking at the intricacies of the problem.Social mobility is one such important tool which helps in reaching the cause of the problem andfocuses on bringing long term equality in the country. The purpose of this study is to examinethe role of social background and education attainment in generating occupation mobility in thecountry. By applying an extended version of the RC association model to 68th round (2011-12) ofthe Employment and Unemployment Survey by the National Sample Survey Office of India, we foundthat the role of education is not important in generating occupation mobility in India, while socialbackground plays a critical role in determining one’s occupation. This study successfully highlightsthe strong intergenerational occupation immobility in the country and also the need to focus oneducation. In this regard, further studies are needed to uncover other crucial factors limiting thegrowth of individuals in the country.
Keywords:
Social mobility, RC association models.
The notion of social mobility is related to equality of opportunities so that individuals can achieve highersocial position regardless of the social background of their parents. It has two motivations, first, byallowing a better utilization of available talents it leads to increased overall efficiency and productivity inthe labour market; second, its objective seems more realistic than equality of outcomes among citizens,which is a desirable objective under many point of view, (Corak, 2020). It encourages human capitalinvestment that can be made equally available to all sections of society through better public institutionsand its policies. While equality of opportunities leads to more social mobility, higher income inequalitythreatens social mobility. In this context, the famous Great Gatsby Curve shows negative cross-countryrelationship between income inequality and inter-generational mobility mentioned in Corak (2013); whichsuggests that inequality skews opportunity and lowers inter-generational mobility.In ancient India, education, skills and occupation were determined by the caste of a person, thus therewas not much freedom for moving between different levels of society (Deshpande, 2010). Although, since1950, the emphasis was on abolishing the caste structure and providing equal opportunities to all, butstrong limitations still exist in the country’s occupational structure as shown by Reddy (2015). Withinthe same period, the country has experienced a substantial increase in income inequality which can beproved by the fact that the share of the top ten percent income group in national income is increasing andthe share of middle 40 percent and lower 50 percent income groups is decreasing (Chancel and Piketty,2019). Interestingly, during this period, the country has also experienced rapid economic growth. In thisregard, Aiyar and Ebeke (2020) concludes that the low level of inter-generational mobility may be thecause why high economic growth coexists with rising income inequality.If we consider this to be a meaningful explanation in the case of India, then it would be interesting tostudy more in depth of occupational intergenerational mobility to get a better understanding about thecurrent situation in the country. As education is considered to be directly associated with occupation, if wetake education and occupation together, it is possible to realize whether education supports occupationalmobility. If it is not supported with the attainment of education, then a conclusion can be drawn aboutthe direct transmission of occupation which mostly goes against the idea of equality of opportunities.1hus, social mobility in this study includes measurement of occupational inter-generational mobility asin Erikson and Goldthorpe (2002) and social background is measured by the occupation of the individual’sparent.In this regard, the present paper attempts to examine the three principal research questions in thearea of social mobility which are: (i) Do mostly sons of fathers with high level of occupation get highereducation? (ii) Do mostly sons with higher education enter a higher level of occupation? (iii) on thewhole, how strong is the association between the occupation of fathers and sons ? The purpose of thispaper is to look at the current occupational immobility by associating it with educational attainmentand social background. So, this investigation is based on the assumption that occupational mobilitydepends on educational attainment and social background. We use 68th round of NSSO data for ourstudy which has been extensively used to study intergenerational mobility. By using an extended versionof the Row-Column (RC) association models, which has never been applied before within the mobilityfield, we expect to complement the existing literature.We find that the association of an individual’s social background with his education is moderate, whilethis relationship is quite strong with occupation. This is because, according to our results, educationdoes not seems to play a huge role in deciding one’s occupation in India. These findings are consistentwith existing literature and it emphasizes the lack of quality education in the country. The rest of thepaper is organized as follows: section 2 reviews the existing literature on inter-generational education andoccupational mobility, section 3 deals with description of the data and socio-economic characteristics ofthe working sample, section 4 discusses association method and section 5 presents results and its analysisfollowed by the discussion and conclusions in section 6.
The first area of study relevant for this paper is human capital theory which was developed by Becker andMincer and focuses on parents’ decision to invest in children’s education and its impact on their incomeand occupation levels Becker and Tomes (1979). Parents investing on the education of their children maybe seen as a way to affect the occupation they may obtain by investing to provide them with better skillsand knowledge.Status attainment theory focuses on additional factors, above and beyond the level of schooling, bywhich parents transfer, by family interactions, life styles and other advantages to their children thatpersist throughout life, including prospective adult wage advantages (Haveman and Wolfe, 1995). It maywork by direct transfer of benefits from parents to their children if, for example, the son of a father witha better profession may get the same occupation due to family ties.Next, Weber’s concept of social closure discusses how ”social collectives seek to achieve maximumrewards by limiting access to resources and opportunities to a limited circle of eligible” (Parkin, 1979).For example, in order to get admission in good universities, if a person needs certain qualities, whichare generally available among children from affluent backgrounds, then it will prove to be an obstacle forchildren with less fortunate background to get admission in such universities (Fishkin, 2012).Concerning empirical studies, we now review some applications with reference to the situation inIndia. Using National Election Study (NES) data of 1996, Kumar et al. (2002) described occupationmobility in terms of origin and destination. They found that 90 percent of the people in farming camefrom farming background which may be due to transfer of land from father to his son. Salary class (whichusually consists of white collar and skilled occupations), apparently reach their position starting fromfathers of diverse backgrounds. Also, 68 percent individuals from unskilled background remain unskilled.Along the same line, Motiram and Singh (2012) using the first round of India Human DevelopmentSurvey of India (IHDS-1), showed that mostly the sons of unskilled and low paid fathers remain in thesame occupation. Another study on education and occupation inter-generational mobility using NationalSample Survey Office (NSSO) rounds from 1983 to 2005 has shown convergence in rates of conditionalprobabilities of education mobility among non SC/STs and SC/STs caste groups (Hnatkovska et al.,2013), which suggests that differences in rates of mobility between these two groups have reduced, however,when it comes to occupational mobility, stagnation still exists which is due to factors other than caste. TheScheduled Castes (SCs) and Scheduled Tribes (STs) are among the most disadvantaged socio-economicgroups in India. Hnatkovska et al. (2013) used median wages to classify occupations, and EUS datausually has many missing values in wages and incomes, mainly for self-employed farmers whose proportionis large in rural India. Next, they kept grandfather and father in the same generation and the child and2he grand child together in the next generation which is usually not appropriate when we want to explorethe mobility between adjacent generations. Further, they used regression and transition matrices tomeasure the education and occupation mobility. The probit regression, on the one hand, does not takeinto account the distance between the occupations of the father and son and only observes whether theson leaves the father occupation and the transition matrix only shows the distribution pattern.Reddy (2015) measures changes in the occupational mobility using the same data up to the year2011-12. In this, the author suggested, there exists less occupational inter-generational mobility in India,especially among the Scheduled Castes (SCs) and Scheduled Tribes (STs). We note that the methodused in the above study is complex, involving few steps that can be avoided if using log linear or relatedinteractions which are not affected by changes in the marginal distributions. It is useful to mention thatthe interaction parameters in the RC model are not affected by the marginal distribution, so there isno need of standardizing the mobility tables required to have the same occupational distributions as inReddy (2015).With regard to education mobility, Kishan et al. (2018), by looking at the correlation between fatherand sons’ years of schooling, suggest education mobility. On the same line, Ray and Majumder (2010),using the 1993 and 2004 NSSO rounds, suggested less mobility for both occupation and education, withoccupational mobility being less than education mobility. Next, Azam and Bhatt (2015), using the firstround of the IHDS data, estimated average inter-generational correlation for India at 0.523 which is higherthan the average global correlation of 0.420. Also, they suggested strong association between expenditureon education with the estimated inter-generational mobility in education attainment.Mueller (2000) compared association between occupation and education mobility between the UnitedStates and Germany using the International Social Survey of Program (ISSP) 1987 for Germany andGeneral Social Survey 1994 for the US. The author finds that social origin have a strong ties witheducation attainment which is associated with later access to occupation opportunities. For instance,higher education has strong ties with white-collar occupations. In comparison, Germany has been shownto have more mobility than the United States. Meyer et al. (1979), compared occupation and educationmobility between Polish men and American men using regression analysis on the 1972 and 1976 surveydata sets. They also suggested that the type of school determines occupational attainment. Further,Carnevale et al. (2011) used the American Community Survey 2007-09 to predict higher education opensup access to higher paid jobs through the use of synthetic estimates of work life earnings. Finally, wewere unable to find much studies on the association of education with occupation mobility in the Indiancontext. In addition, the use of RC models has been more recent in this area through the use of mobilitytables, which we expect will strengthen the existing literature.
The data used in this paper come from the 68th round (2011-12) of the Employment and UnemploymentSurvey (EUS) conducted by the National Sample Survey Office (NSSO) of India. The EUS providesprimary source of data for various indicators of labour force at state and national level. It followsa stratified multi-stage sample design and includes a sample of around 100,000 households coveringalmost all geographical regions of the country. It is the largest data gathering information on almostevery social and economic aspect at the individual and household level since 1983 in India. It containsinformation about education in 13 broad categories ranging from not literate to graduate and above andoccupation levels are classified according to the national classification of occupations (NCO-2004) four-digit occupation codes. The basis of divisions in the occupational structure is based on the skills requiredto perform the functions and duties of an occupation.Initially we arranged the education categories into six groups: not literate, without formal schooling,primary, secondary, higher secondary or diploma certificate and graduate and above that ranged from1 to 6, respectively. However, because the proportion of sons in the second category of education isless than 0.2 percent in our sample, we decided to merge categories 1 and 2, thus, in the analysis,education is taken as having 5 categories. We categorized occupation codes into four categories asunskilled, farming, skilled/semi-skilled and white collar respectively by following the NCO single-digitoccupation codes of Labour and Employment (2004) and Reddy (2015) occupational structure. It isworth noting that there is no uniformity in selecting the framework of occupational structure as literatureexists with different structural frameworks by different authors in the context of the same country.Here, the unskilled occupation includes labours from agriculture and fisheries, mining and constructionactivities. The farming business includes market oriented skilled and subsistence agriculture and fisheryworkers. Skilled and semi-skilled occupations include office clerks, service workers, sales workers, craft-3elated trades workers, plant and machine operators, and assemblers. White collar occupations includelegislators, managers, professionals.The NSSO data does not contain information about parents if the person is living separately from hisfamily. Therefore, in order to do study on inter-generational mobility, we selected only those householdswhere the working person and his father are living together. Also, we concentrate on male subjectsbecause married women in India live with their husbands or father-in-law and the survey does notprovide information on their parents. Thus, the criteria for selecting the working sample were householdswhere the son’s age was between 16 and 45 and both father and son were not currently enrolled in anyeducational institution and informed about their education and occupation. The above criteria for sampleselection provides a sample of working father and son from which we removed cases where the requiredinformation was missing. This procedure lead to a sample of 27771 households which is our ’workingsample’. In case a father was living with more than one working age son, we selected only the eldest sonto ensure that we are obtaining the record of a father and a son in our working sample.To check whether the selection leading to our working sample is unbiased, we compared the socio-economic characteristics of co-resident sons with sons who are living separately from their fathers. Inpractice, non co-resident sons correspond to households with only one adult male who is of working age.We found 48390 non-co-resident households in our sample.Table 1: Summary statistics for sons who are co-resident or are living on their ownco-resident Living on their ownVariable Obs Mean SD Obs Mean SDAge 27771 25.91 6.12 48390 35.83 6.54% of Rural Pop. . . . . . . . . . . . . . . . . . . . . . . . .
Statistical methods suitable for the analysis of social mobility depend both on the nature of the data andon the purpose of the analysis. For instance, when, like in Mazumder (2016), one has income data atthe individual level for the father and the son, methods based on linear regression on incomes or on thecorresponding ranks may be used, depending on whether one believes that the relation is approximatelylinear or not. Instead, when, like in our case, data are in the form of contingency tables, methods basedon interactions are more suitable. Another important distinction is whether one aims to summarize theoverall degree of association by a single number like in Altham and Ferrie (2007) or to undertake a moreanalytical investigation, looking at several measures of association at the same time.There is substantial agreement in the literature that the set of log-linear interactions computed ona contingency table provide one of the best assessment of the strength and the direction of associationbetween the row and column variable. Clearly, stronger association means that the social class of the sonmay be more easily predicted from that of the father, thus, stronger association is equivalent to smallerchances of social mobility. An important property of interaction parameters is that they are not affectedby the structure of marginal distribution. This is related to the algorithm described in Altham and Ferrie(2007) which allows to transform a given contingency table into another having the same set of interactionsand arbitrary marginal distributions. This may be important in the light of separating structural fromrelative or circulation mobility as discussed, for instance by Hauser and Grusky (1988) and Sobel et al.(1985).It is well known that in an r × c contingency table, we can compute ( r − c −
1) non redundant log-linear interactions measuring the degree of immobility within different subsections of the table. Thereare, essentially, two different strategies to deal with such a multitude of measure: (i) to compute aunique summary measure by some appropriate average as in Altham and Ferrie (2007), an approachapplied, for instance, in Reddy (2015), or (ii) try to fit some restricted model depending on a smallernumber of parameters, a route followed in this paper where RC association models are applied. RCassociation models were introduced by Goodman (1981) to simplify the association structure withoutloosing important information. These models have been used for the analysis of social mobility by, forinstance, Xie (1992) and Mueller (2000). An RC(1) association model has just one coefficient of intrinsicassociation: higher values of this coefficient indicate stronger association and thus lower mobility. Inaddition, the estimated model provides a set of row and column scores from which we can measurethe relative distance between categories: if two categories are close to each other, the correspondingconditional distributions are very similar.Various extensions of log-linear interactions have been studied in order to capture more specific fea-tures of association; they are essentially based on assigning a logit of type L (local), G (global) or C(continuation) to the row and the column variables. A wide collection of interaction parameters ob-5ained by combining different row and column logit types are studied in Douglas et al. (1990) in thecontext of positive association, a notion closely related to social mobility when father and son social classmay be ordered from lowest to highest, in that case, stronger positive association means lower mobility.Douglas et al. (1990) also provide a graphical interpretation of the different interaction parameters. RCassociation models may be used to extract the most relevant features of the association structure in asocial mobility table when interactions are defined by combining row and column logit types, see forinstance Bartolucci and Forcina (2002). One further extension, introduced by Kateri and Papaioannou(1994), has allowed to combine traditional RC association models, Correspondence analysis and a wholecollection of other models into a unified class of RC association models depending on a scaling factor.The statistical methods used in this paper are based on the even larger class of RC association modelsof Forcina and Kateri (2020) which allow the user to choose both the type of interaction parameters asin Douglas et al. (1990) and the scaling factor as in Kateri and Papaioannou (1994). The advantage ofthis approach is that we may easily explore a large range of different models and select the one that isas simple as possible and fits the data best. The strategy used in this paper is to search for the smallest K such that an RC ( K ) model fits the data sufficiently well. For the three tables analysed in this paper,no satisfactory model with K = 1 seemed to be adequate; on the other hand, it was possible to find an RC (2) model which fits the data very accurately. While the deviance is uniquely defined, computationsof the coefficients of intrinsic associations and the rows and columns scores depend on row and columnweights; we adopted the usual strategy (see Kateri, 2014, Chap. 6) based on uniform weights.The strength of immobility in an RC(2) model depends on two coefficients of intrinsic association,where higher association means more immobility. To give an idea of the degree of immobility impliedby a given pair of coefficients, below we compare several hypothetical version of the association betweenfather occupation and son education. More precisely, we consider the joint frequencies that we had got if,keeping the rows and columns score fixed to the vales estimated by the best model, the pair of coefficientsof intrinsic association, relative to the values estimated in the best fitted model were: a - the same, b -both divided by two, c - both multiplied by 2.5.Table 4: Theoretical joint frequencies for the education of sons of fathers in U and W in three hypotheticalscenarios Son EducationFather N P S H GU, a 525 1221 1711 304 103U, b 381 898 1639 550 396U, c 1325 1690 756 77 16W, a 134 499 1691 1073 1377W, b 207 707 2108 943 809W, c 0 13 1047 1609 2106 We now study the joint distribution of father’s occupation and son’s educational attainment in India.This will help us understand to what extent educational attainments of the son depends on his father’soccupation in the sense that father with a better occupation have better chances to invest more in theeducation of their sons.Table 5: Observed joint distribution of households by father occupation and son educationSon EducationFather Occ N P S H GU 526 1222 1707 307 102F 731 1911 5046 1916 1224S 512 1664 3742 1370 1017W 135 501 1686 1076 13766t first, a collection of extended RC(1) models as in Forcina and Kateri (2020) were fitted by settinglogit type for occupation to L because its categories are not necessarily ordered and L, G and C foreducation, for a range of values of the λ parameter; the best of these models had deviance of about 28 on6 degrees of freedom, which is significant. Thus, we moved to RC(2) models: the best fit was obtained bysetting logits to L for occupation and G for education with λ = 0 .
22. This model has a deviance of 0.14on 2 degrees of freedom. The coefficients of intrinsic association are equal to 0.99 and 0.02 respectively.The row and column scores are plotted in Figure 1 -2 -1.5 -1 -0.5 0 0.5 1 1.5-2-1.5-1-0.500.511.52 U FS WN P S H G
Figure 1:
Plot of row (squares) and column (diamonds) scores for the data in Table 5 .It is interesting to note that on the horizontal axis, which corresponds to the largest coefficient,education categories are ordered and almost equally spaced; the same is true for occupation, but F and Sare very close to each other on the horizontal axis. Note, also that the pairs (U,N) and (W,G) are bothclose suggesting that the sons of unskilled workers are the most likely to achieve no formal education whilethose of white collars are the most likely to get a G+ degree. However, from the frequency distributionin table 5, we see that about 75% of the sons of unskilled father get primary or secondary education.Probably, this is the result of schemes like Sarva Shiksha Abhiyan (SSA), Mid-Day Meal Scheme, Right toEducation (RTE) Act which have helped children from poor backgrounds get enrolment up to secondarylevel. On the horizontal axis S (education) is very close to both F and S, indicating that a large proportionof sons of farmers or skilled-semi skilled workers get secondary education. On the whole, considering alsothe coefficients of intrinsic association, we may say that the effect of father occupation on son educationis active but to a moderate degree, allowing for a reasonable amount of mobility. However, much remainsto be done to enable and cover such students in the workforce.
The purpose of the following analysis is to determine how much the the efforts spent in getting a bettereducation improve the chances of getting a better job, in other words we examine the role of education inachieving higher level jobs in India. It is worth noting that here strong association means, roughly, thatpeople get the job for which they are qualified, instead, weak association indicates that other factors, likefamily influence and connections, play an important role.Some preliminary model selection suggested that no RC(1) model fits sufficiently well the data, so weexamined a range of RC(2) models, the one with logit type C for education and L for occupation with λ = -0.06 fits best with a deviance of 0.11, which means an almost perfect fit. The estimated coefficientsof intrinsic association are equal to 1.03 and 0.02 respectively. The row and column scores are plotted inFigure 2The plots in Figure 2 indicate, again, that, on the horizontal axis, education categories follow thenatural order and are almost equally spaced. Occupational categories follow also the expected order onthe horizontal axis, except that F and S are very close to each other, though they diverge on oppositedirections on the vertical axis. The fact that G and W are very close to each other, means that sons witha G+ degree have the highest chances of becoming white collars. The same is true for the pair N and U,7able 6: Joint distribution of households by son education and son occupationSon occupationSon Education U F S WN 655 579 572 98P 1544 1461 1933 360S 2134 3996 4739 1312H 363 1466 1749 1091G 100 778 1057 1784 -1.5 -1 -0.5 0 0.5 1 1.5-2-1.5-1-0.500.511.52 N P S H GU F S W Figure 2:
Plot of row (squares) and column (diamonds) scores for the data in Table 6 .but only on the horizontal axis: the most likely occupation for sons with N (education) is U, but theyhave non negligible chances of ending up into F or S. On the horizontal axis, S (education) is betweenF and S (occupations), meaning that sons with secondary degree are most likely to become farmers orskilled workers. On the whole, the strength of association is only a little stronger than in the previoustable, meaning that education is not the only factor that determines the kind of occupation that a personcan acquire.
The purpose of the following analysis is to examine the shape and strength of association between fatheroccupation and son occupation. This is important to answer the following question: the effect of father’soccupation on son’s occupation is only indirect, that is induced by the fact that fathers with a betteroccupation can afford to invest more to provide a better education to their sons who, because of theireducation, can get a better job, or there is also a direct effect, in the sense that the sons of fatherswith a better occupation, because of family ties, can get a similar occupation even if not adequatelyqualified. For these data all RC(1) models fit badly irrespective of the logit types while the RC(2) fitsTable 7: Joint distribution of households by son education and son occupationSon occupationFather occupation U F S WU 2644 192 898 130F 988 6861 1917 1062S 845 730 5952 778W 319 497 1283 2675very well, so we set both logit types to G and searched for the optimal value of λ which equals -1.21with a deviance of 0.01 on 1 degree of freedom. The two coefficients of intrinsic association equal to 3.35and 0.24 respectively, almost three times larger than in the previous two cases above, indicating that,8robably, family ties must be operating in addition to education.Both the rows and columns scores follow the natural order on the horizontal axis which is the mostimportant. Note also that each category of father occupation is fairly close to the corresponding categoryof son occupation, at least on the horizontal axis, which suggests that, to some degree, sons tend toremain in the same occupation of their father; indeed, the largest frequencies are along the main diagonalin Table 7 -1.5 -1 -0.5 0 0.5 1 1.5-2-1.5-1-0.500.511.52 U F S WU F S W Figure 3:
Plot of row (square) and column (diamonds) scores for association between father and sonoccupations in Table 7 .The above analysis shows that the association of father occupation to son occupation is strong.This implies that regardless of a person’s education background, a son is more likely to get the sameoccupation of his father. Thus, it can be concluded that the connection is direct rather than mediatedthrough education. If we try to match the ground reality with our results, then our results match thepractical aspect prevailing in India. In India, it is found to a large extent that the father tries to keephis child in his profession. This may be due to less return from education as in Shrivastava et al. (2019);Aggarwal (2014) and hence father’s influence in the labour market predominates in deciding his child’sprofession. This is consistent with the inference that wherever there is less return from education andskills, occupation pathway becomes the primary channel of inter-generational persistence (Blanden et al.,2014).
In this paper we have investigated inter-generation social mobility in India by using the 68th round ofNSSO data for 2011-12 year. Our results indicate that the association between father occupation andson educational attainments is moderate, meaning that, probably because of the present policies of thegovernment, together family efforts, the sons coming from a modest background have over 50% chancesto reach, at least, secondary education. Unfortunately, the association between son education and sonoccupation is also moderate, indicating that education is not the main factor that determines occupationand, thus, social position. This finding is confirmed by the fact that the association between father andson occupation is much stronger than those passing through education. This means that there are otherfactors that determine one’s occupation apart from education. Overall, it suggests that the role of socialbackground in deciding one’s education is only moderate while the role of the same social background isstrong for deciding one’s occupation. The strong dependence of occupation on social background suggeststhat India is still not an open society and especially opportunities for work are not quite distributed.We believe that there are three important interpretations for the above paradigms of social mobility inIndia. First, India’s social structure evolved from a rigid caste structure but still there exist restrictionsin society especially at the lower level, which do not allow certain groups to grow and take advantageof development. Second, the limited role of education in determining one’s occupation also exists dueto unsatisfactory quality of education in the country. This is proved by the fact that, despite severalinitiatives taken by the government at the lower level of education, only 9 out of 28 states have shownimprovement in the School Education Quality Index (SEQI, 2019), while for 9 states it has gone down and9he rest show no change as per National Institution for Transforming India (NITI Ayog). Further, if welook at India’s position in advanced education, its score is 56.42 which is one of India’s lowest componentscores in the Social Progress Index (SPI 2020). At the same time, if we look at the component score forthe quality of education of Scandinavian countries, it is quite higher than many countries in the world.Overall, their ranking in the Global Social Mobility Index 2020 and SPI 2020 is quite high and the rateof inequality is also very low in these countries. Thus, it is possible to say that social mobility, whichhas been seen as an important tool to bring long term equality, has a clear link with fair education andoccupational opportunities in the country. Third, other important factors such as health, infrastructureand technology are currently under development in the country, which directly contribute to the abovesocial mobility indicators. Since India’s resources are diverse and the requirements of one state may bedifferent from others, a state-level study on social mobility indicators at the national level will help identifythe lack of components at the national level and demonstrate the need for immediate improvement atthe regional level. We intend to study social mobility indicators at the state level in subsequent work.