Measuring Inequality of Opportunity in Pakistan: Parametric and Non-parametric Analysis


Abstract. The present study estimates inequality of opportunity for Pakistan by using parametric and non-parametric analysis. Pakistan Social and Living Standard Measurement Surveys of two time periods, 2005-06 and 2010-11, have been utilized. Father’s education, mother’s education, father’s occupation, region of residence, and gender have been used as the circumstance variable while labour earnings and household income per capita are outcome variables. Three indices of Generalized Entropy measures have been utilized to estimate inequality of opportunity. Results depict that inequality of opportunity in labour earnings declines by 11 percentage points in Pakistan, during study period. The inequality of opportunity, for household income per capita, declines by 16 percentage points for Pakistan. Among all circumstances, gender is the highest contributor followed by region of residence, father’s education, and father’s occupation.

Keywords: Inequality of opportunity, Parametric analysis, Non-parametric analysis, Pakistan


    Since last two decades, a huge literature on the measurement of inequality has been emerged. Broadly, this literature can be classified into two types: measurement of inequality, and determinants/causes of inequality. Among the determinants/causes of inequality, the impact of economic development on inequality is the most debatable issue after the work done by Kuznets (1955). The inequality-growth relationship suggests several channels through which inequality can affect growth. Unobservable efforts (Mirrless, 1971), accumulation of saving (Galenson and Leibenstein, 1955), and size of investment projects (Barro, 2000) are amongst the important paths through which equality may enhance growth (Marrero and Rodríguez, 2012).

    Contrary, political instability (Alesina and Perotti, 1996), capital market imperfections (Banerjee and Newman, 1991), unproductive investment (Mason, 1988), and regional disparity (Sandilah and Yasin, 2011) are amongst main reasons of negative relation between inequality and economic growth. Depending upon the dominance of the above mentioned channels, the inequality can be negatively or positively affected by economic growth.

    The impact of inequality on growth turns out to be ambiguous because several studies investigated the positive relation while some explores the negative (Marrero and Rodríguez, 2012). Studies highlighted that this might be due to usage to contradictory inequality indices (Knowles, 2001; Szekely, 2003) or the inconsistent econometric methods (Forbes, 2000). This ambiguity can be due to the indistinct conceptual understanding of inequality (Marrero and Rodríguez, 2012). The policy makers and the government adopt several measures to remove inequality without the clear understanding of its dimensions. For the minimization of inequality it is very much essential to target the right path so that level playing field is provided for all individual. After the seminal work by Roemer (2000), it has been accepted that fraction of inequality is linked to the differences in circumstances and opportunities faced by individual (Hassine, 2009).

    Inequality of opportunity (IO) refers to that inequality that is due to circumstances, which are beyond the control of individual, and effort. The circumstances are those factors which an individual receives by the day of birth, e.g. family background, region of residence, father’s education, race etc. The second factor effort, is entirely linked towards individual’s choice, e.g. number of hours worked, occupational choice, migration etc. Consequently, the overall inequality is the result of heterogeneity in social origins and other factors such as effort (Marrero and Rodríguez, 2012). Coming to our starting point, whether inequality has positive or negative relation with growth, the two fractions of inequality have different effect on growth. The provision of equal circumstances contributes a larger portion in the enhancement of growth rather than effort, because circumstances play dual role. Firstly, circumstances directly minimize the outcome inequality.

    Secondly, also play a role for the minimization of outcome inequality through effort (de Barros et al., 2009). Hence, the success of policy interventions in alleviating inequalities and improving welfare depends upon their efficacy in compensating for the circumstance-based disadvantages and in expanding opportunities (Peragine, 2004; Ferreira and Gignoux, 2008).

    Several studies estimated inequality of opportunity for different countries; I have not come across any study which does it for Pakistan. IO to be a useful for policy, we need to estimate of how much of total inequality is due to IO. Energies should aim to get rid of that part of inequality which is due to circumstances, which are beyond the control of an individual. Given this context, the present study estimated the inequality of opportunity for labour earnings and household income per capita in case of Pakistan. By using PSLM of two time periods, 2005-06 and 2010-11, study estimated IO by utilizing parametric and non-parametric techniques.

    The rest of the study is organized as: section II discusses the studies relevant to inequality of opportunity, section III explains the parametric and non-parametric techniques to estimate IO. Section IV presents the results of parametric and non-parametric analysis and last section concludes the study and recommends certain policies to minimize IO in Pakistan.


    There is immense literature available on this topic and several researchers estimated IO. Most of the studies are found in case of Brazil and EU countries. Marrero and Rodrí´guez (2012) have tested whether the effects of macroeconomic changes on income inequality depend on their influences on the total inequality constituents, i.e. inequality of opportunity (IO) and inequality of effort (IE). The research uses US data from PSID for a period of 1970-2009. A dynamic model was used to relate the components of inequality with macroeconomic factors as real GDP, inflation rates, outstanding consumer credits, public welfare and health care expenditures. The results revealed significant negative effect of real GDP and outstanding credits upon IO and IE, while positive significant effect of inflation only on IE, and welfare expenditures only on IO.

    Gamboa and Waltenberg (2012) used PISA data of 2006 and 2009 to detect the IO for educational achievement in six Latin American Countries using non-parametric approach. School type, gender, parental education and their combinations are taken as circumstance variables. Countries are ranked according to the degree of IO prevailing in them on the basis of decomposable inequality index. The unconditional inequalities based rankings showed limited similarity with the conditional inequalities based rankings. Gross inequality is decomposed into opportunity and effort fraction. Overall, gender based IO was insignificant except for reading whereas school type based IO was highly significant. The index of IO ranged from below 1% to 27%, varying significantly across years, countries, subjects and circumstances. Bootstrap and alternative index are used to verify the results.

    Marrero and Rodrí´guez (2012) explored the reasons for the inconclusive effects of income inequality on growth. They postulated one potential reason for this confusion to be the two contradictory components of income inequality, i.e. IO and IE. These components affect growth in opposite directions using means of min approach and hence, the direction of relationship remains subject to the direction of the dominant force. Separate analysis for IO and IE was taken up using the PSID database for 23 states of the US in 1980 and 1990. A negative relationship was observed between IO and growth, and a positive one between IE and growth.

    Dabalen et al. (2015) conducted a study to inquire the IO among Egyptian Children in access to basic needs. It analyzed the degree of EO regarding basic services taking gender, birth place and family background as circumstance variables and health, access to basic services and income as outcome variables. Data for 2000 and 2008 were taken from Egypt Demo-graphic and Health Survey (DHS) the Egypt Household Income, Expenditure and Consumption Survey (HIECS). Results for IO reveal that although reduced it is still significant in case of school enrollment and healthcare. IO is insignificant for malnutrition, safe drinking water availability, electricity, sanitation etc. The study thereby reveals that Egypt has attained significant progress with reference to the availability of and access to basic services for children and mothers, in some cases with a pro-poor overall effect.

    Abras et al. (2013) endeavored to quantify the degree of IO in labour market for selected European and Central Asian (ECA) countries, to compare across ECA countries and with self-assessments and to compare ECA results with Latin American and Caribbean (LAC) countries. Decomposing inequalities into circumstance variables (gender, parental education, minority status etc.) and effort variables (education, age), Human Opportunity Index and Theil-L index were calculated using data from the Life in Transition Surveys (LiTS) 2006. The findings showed significant circumstance based IO in employment status for ECA region and high heterogeneity across countries. Significant positive correlation was observed between the Theil-L index and the self-perception of equality. The comparison suggested that ECA countries do much better than the LAC countries.

    Dabalen et al. (2015) conducted a study on economic situation, especially IO in South Africa. HOI and Gini coefficients were used for measuring IO. After the global economic crisis, South Africa’s economy has yet been unable to gain momentum. Industrial concentration, labour market rigidities, skill shortages and low savings and investment rates...

To continue reading

Request your trial