Author: Sujan (page 1 of 2)

The Whole Nine Months: Fertility Rates as Predictors of Economic Growth

[This article was published by TheWire.in on 9th April 2018. It was written in collaboration with Aniruddha Ghosh, a classmate from LSE. To read the article on TheWire.in, please click here 

Featured on the Harvard Economics Review, click here ]

 

Nearly 10 years after the Great Recession of 2007-09 that brought monetary and real disturbances across the world economy, researchers have stumbled upon a new and a rather ingenious business cycle fact that may aid in predicting upturns and downturns in an economy: the growth rate of conceptions as predictors of economic growth.

While economists have uncovered such a relationship for the United States, a preliminary reading of Indian data also suggests similar patterns. Of course, a more in-depth and rigorous analysis is needed to unmask the trends. Such trends could be an aid in augmenting the much needed high-frequency indicators to aid policy making for the Indian economy.

It is a well-known fact that modern developed economies, as documented by most recent studies, have tended to show a positive association between fertility rates and economic growth: procyclical fertility. Patrick Galloway in his 1988 study on pre-industrial Europe highlighted the sharp responsiveness of fertility rates to the price of food grains. A higher price of food grains meant lower economic well-being, in turn negatively influencing fertility rates.

Moreover, the magnitude of positive association between fertility rates and economic growth was consistent across most of the pre-industrial European economies. While the fact remains that cross-country studies report a negative relationship between per-capita income and fertility levels, fertility rates indeed tend to be procyclical with economic movements. Indeed, statistics from the US’s National Center for Health Statistics show how the number of births began to decline in 2008 as the fallout of the Great Recession became prominent, falling by over 15% after hitting a 50-year high in 2007.
In addition to these facts, economists have deciphered a new fact about the movement of fertility rates. According to a recently published working paper at the National Bureau of Economic Research (NBER) – a leading economic research organisation in the US – fertility data not only moves in tandem with economic cycles but can even help predict the onset of a recession or an upturn. Studying quarterly economic and demographic data for the last three decades (1988-2015) , researchers Kansey Buckles and Daniel Hungerman from University of Notre Dame, and Steven Lugauer from University of Kentucky find that the growth rate of conceptions falls sharply during a recession, and importantly, this fall actually begins a few quarters before the onset of a recession.

In the jargon of macro-economics, the growth rate of conception is a leading indicator of business cycle, having predictive power in foretelling an impending downturn. Technically, conception is a measure based on births, therefore, it has a built-in nine month lag attached to it.

Figure 1: Growth Rate of Conceptions fall sharply during recessions, and this fall typically begins a few quarters before the onset of a recession (US 1988 Q1:2015 Q4). Source: Buckles, Hungerman, and Lugauer (2017)

This anticipatory nature of the growth rate of conceptions has been observed for all the major recessions in the last three decades, that of 1990, 2001, and 2008, and has been more pronounced in the recent recession. This high-frequency conception data allows authors to state a rather unconventional business cycle fact and we quote, “the growth rate of fertility declines prior to economic downturns and the decline occurs several quarters before recessions begin.” It is important to keep in mind that the fertility decline is solely driven by fall in conceptions for the US data rather than abortions or foetal deaths. The extent of fall in conceptions is large, growth rate of conceptions fell nearly 5 percentage points at the onset of Great Recession of 2007-08.

As mentioned above, the findings show that the growth rate of conceptions typically starts falling a few quarters before a recession, and since its is measure based on births, has an implicit 9 month lag built in to it. So what explains these possible associations? The hypothesis that researchers put forward is a straightforward one: Couples make forward looking decisions on having children based on economic outlook. If this hypothesis is true, then fertility behaviour in the US over the last three decades has been forward-looking and quick to respond to changes in the state of the economy. The explanation does depend on the assumption that almost all the conceptions are ‘planned’, i.e. couples are making conscious choices about bearing children, as opposed to unplanned, i.e. children born due to lack of contraception, out of wedlock, etc. With unplanned pregnancies being quite high in the US (estimates put it between 40-50% of all pregnancies), there is a legitimate question as to whether there are other factors in play which are causing this predictive relationship between growth rate of conceptions and economic growth.

Nevertheless, if future economic conditions matter for current conception decisions, and if expectations are at least rational or future incorporating, movements in current conceptions may likely be harbingers of future conditions.

Figure 2: Growth Rate of Conceptions and economic growth during the 2007-09 Great Recession. Source: Buckles, Hungerman, and Lugauer (2017)

That’s one part of the story. Interestingly, the researchers also find that the growth rate of conceptions tends to moves together with common economic indicators like consumer confidence and durables purchases, If you look at Figure 3, the fall in growth rate of conceptions coincides with or even precedes falls in consumer confidence and durable purchases. Consumer Confidence Index, Durable purchases are often looked upon as high frequency indicators that can foretell the prospects of the economy in the near term. In India, the Reserve Bank of India (RBI), Centre for Monitoring Indian Economy (CMIE) track independently consumer confidence which are useful aids in capturing expectations for the near term future of the economic situation.

Figure 3: Fall in conventional economic indicators like consumer confidence and purchase of durables, and the growth rate of conceptions tend to be coincide during recessions. Source: Buckles, Hungerman, and Lugauer (2017)

What does the trend seem like for India? We do a very preliminary exercise here and graph the growth of birth rate and GDP series for India. The graph points to some predictive behaviour but of course, this has to be subjected to more stringent and robust econometric evaluations. Also, we use birth rates, a measure different from conception rates that Buckles et.al have used. However, as a first pass, this must interest practitioners to look carefully between the associations. Although, we think that the hypothesis that Buckles et.al put forward may not hold tightly for India. With a low per capita income, a gloomy economic situation might step up conceptions as one needs more hands to feed. Again, all these are possible speculations and a careful analysis of the data will only uncover the mechanisms in place.

See Figure 4 here.

While whether conception rates or birth rates should be the variable of interest can be discussed, a common challenge with using both these measures is that they are hard to measure in real time, even when compared to other measures like consumer purchases. However, with the coming of sophisticated data analytics and statistical techniques, it is quite possible to get predictions of these measures on a real time basis. One possible way to track conceptions is to use purchases of goods that are especially likely to be bought by those who are attempting to conceive or who are newly pregnant. This data is usually tracked real time by retail firms using scanner technology. In fact, this data is already being used by retail firms for targeted marketing and advertising.

This New York Times piece in the year 2012 outlined how Target Corporation could observe, track and study consumer purchases and predict if a customer is pregnant. Most of the online portals place cookies on your devices to track your shopping habits. Therefore, there are tools that are available to harness real time data on conceptions. Of course, such tracking has to respect the contours of privacy since these are very private issues Buckles et. al explores few choices on tracking conceptions and interested readers are suggested to have a look at the appropriate sections.

Yogi Berra (the legendary Yankees Baseball player), famous for his Yogi-isms, had once quipped, “It’s tough to make predictions, especially about the future.” It’s encouraging for policy practitioners to have such an evidence of conception rates as predictors of economic cycles. To our knowledge, this study is one of its kind and should make researchers across the globe investigate for such associations in their respective geographical spaces. Policy institutions in India often cite the lack of high frequency indicators that can be an aid in the policy formulation process. Buckles et.al should encourage them to hunt for innovative indicators and develop tools that can be usefully utilised. With so much jargon and tools being thrown around Big Data, it will be useful if researchers look at such meaningful constructs like Buckles et.al do.

 

 

 

 

How to Split the Pie and Eat it Too! – A Game Theory Exercise

Bargaining is a very important concept in economics, most notably in game theoretical frameworks. A common problem which one comes across is how to split the pie between two players.  The question is as follows:

There are two players; A, and B. They are given by a pie which they must split between themselves. In Period 1, Player A proposes a division, and Player B accepts or rejects this division. If Player B accepts the division proposed by Player A, then the game ends at that time itself and both players walk away with their share of the pie. If Player B rejects, then Player B proposes the division in Period 2. Now Player A has the option of accepting or rejecting the division. Again the same conditions apply, and the game can continue or end in Period 2. However, for each Period the game goes on for, the size of the pie keeps diminishing by a certain amount. There are a certain number of periods the game can go on for, before the value of the pie becomes zero. 

So how should the pie be divided?

This problem is important, both as an introduction to bargaining games, as well as a introduction to thinking about ‘dynamic’ issues in economics — where agents think across time periods (as opposed to static frameworks where agents are only thinking about the current period of time).

 

I have made a small presentation about this problem and how to resolve it. To check it out, click here!

Price Discount Calculator

Most firms offer consumers one of two things during marketing campaigns and sales to boost sales -: i) Price discounts, like a 50% off on prices; ii) Extra Volume, like 50% extra volume for the same price. But, is a 50% increase in quantity equal to a 50% discount on prices? Actually, no! A 50% increase in quantity is only equal to about a 1/3rd or 33.333% discount on prices.

But according to an article in The Economist (2012) (Something Doesn’t Add Up), and research by behavioural scientists (Chen et al (2012)), people fail to realise the difference between the two and often end up viewing them as the same.

One of the researchers, Akshay Rao of the University of Minnesota’s Carlson School of Management, offered his undergraduate students a choice between two a deals when buying loose coffee beans – 33% extra for free, or 33% off on the price. While the discount on the price is the clearly better deal, the students viewed both deals as the same!

For equivalent deals (like a 33.33% price discount vs 50% extra quantity), researchers found that consumers mostly went for the extra quantity (about three quarters of them), even though the deals were numerically equivalent.

This tool helps you to not fall into this numerical blindspot by calculating what price discount (%) corresponds to extra quantity (%) and let’s you make informed decisions on purchases.

Please find below the link for the app:

https://sujanbandyo.shinyapps.io/price_discount_calculator/

Generating OLS Results Manually via R

Statistical softwares and packages have made it extremely easy for people to run regression analyses. Packages like lm in R or the reg command on STATA give quick and well compiled results. With this ease, however, people often don’t know or forget how to actually conduct these analyses manually. In this article, we manually recreate regression results created via the lm package in R.

Please click here to read the article!

ASER 2017 Shows India’s Secondary Education Sector Is Failing to Impart Basic Skills

[This article was published by TheWire.in on 20th January 2018. It was written in collaboration with Aniruddha Ghosh, a classmate from LSE. To read the article on TheWire.in, please click here .]

 

[This article was cited in an article in the Economic and Political Weekly (EPW) journal –
Nawani Disha (2018), Assesing ASER 2017 Reading Between the Lines, Economic & Political Weekly, Vol 53, No 8.]

 

Merely increasing enrolment will not lead to the development of elementary skills that education is supposed to provide.

ASER’s statistics have shown how the ability of grade 8 students has been consistently falling over the years, coinciding with the increase in enrolment rates. Credit: TESS India/Flickr (CC BY-SA 2.0)

ASER’s statistics have shown how the ability of grade 8 students has been consistently falling over the years, coinciding with the increase in enrolment rates. Credit: TESS India/Flickr (CC BY-SA 2.0)

The Annual Status of Education Report (ASER) 2017 ‘Beyond Basics’ was recently released and is unfortunately on expected lines. The report highlights the sorry state of education when it comes to India’s 14-18-year-olds and asks rather uncomfortable questions that policy practitioners must find answers to. In an earlier article, we had outlined some issues that have plagued India’s primary education sector and some anticipations that we had from the ASER 2017 report. Not surprisingly, we see a repetition of the same issues when it comes to India’s secondary education space.

Previous ASER reports observed that despite high enrolment ratios of over 96% in the last eight years in the primary education sector, improvement in reading outcomes and arithmetic ability continues to be low. Moreover, a large proportion of students in both government and private schools continue to be below the ‘Grade level’. Grade level means that a student can deal with what is expected of her in that grade.

The ASER 2017 is targeted to look ‘beyond basics’: the age group between 14-18, primarily those outside the Right to Education ambit and on the verge of entering adulthood. The government’s flagship Rashtriya Madhyamik Shiksha Abhiyan (RMSA) launched in 2009 and re-booted in 2013 as RMSA-Integrated has not been much of a success in India’s secondary education scene.

‘Aspirations’ of young India

What we find as a salient feature of this report is the coverage on the aspirations of young India. Capturing ‘aspirations’ by well-defined metrics is a tough ask, very few datasets across the world capture them with some rigour. The Young Lives Survey in Peru is one example that comes to our mind that captures subjective well-being and is actively used by researchers to gauge aspiration-achievement shortfall and their reasons. While successive governments pride over India’s sizable ‘demographic dividend’, the ASER 2017 points starkly to the basic skill gaps that plague our young population.

Before documenting some necessary takeaways from the ASER 2017 report, it is prudent to mention that the complexity of the data collected makes the national estimates a summation of estimates generated at the district level (24 states, 26 districts, 23,868 households and 28,323 youths).

Like the primary education sector, enrolment rates have also been high and increasing in the secondary education space. The RTE Act covers mandatory and free schooling up until the age of 14, or roughly grade 8. ASER surveys show that enrolment in grade 8 has been steadily increasing from less than 50% in 2005-06 to close to 90% in 2014-15. However, the quality of education remains a concern. ASER’s statistics have shown how the ability of grade 8 students has been consistently falling over the years, coinciding with the increase in enrolment rates.

While grade 8 enrolment has nearly doubled over the last ten years, the proportion of students acquiring base skills has been reducing. In 2014-15 only 44% of grade 8 students could solve a grade 4 level division problem, down from 72% in 2007-08. Similarly, the ability to read grade 2 level texts has fallen from 87% to 75% over the same period. Source: ASER 2017

While grade 8 enrolment has nearly doubled over the last ten years, the proportion of students acquiring base skills has been reducing. In 2014-15 only 44% of grade 8 students could solve a grade 4 level division problem, down from 72% in 2007-08. Similarly, the ability to read grade 2 level texts has fallen from 87% to 75% over the same period. Source: ASER 2017

Enrolment rates after grade 8

Another important issue is analysing how enrolment rates develop after grade 8, or once students are no longer under the purview of the RTE Act. Looking at the 2011-12 grade 8 cohort, the findings show about a one-third decline until grade 12, indicative of a trend of increasing dropout rates after grade 8. The same trend is reflected when enrolment rates are analysed by age, showing a steady increase from age 14-18.

Once students leave the purview of compulsory and free education at age 14 (around grade 8), enrolment rates for further education drop. Source: ASER 2017

The reasons for discontinuing studies vary. Around 25% of the youth who dropped out after grade 8 said they did so due to financial reasons. Worryingly, a large number of students (34% of boys and 19% of girls) said they dropped out due to lack of interest, pointing to deficiencies in the curriculum and teaching infrastructure. One-third of girl students said they dropped out due to ‘family constraints’.


Also read: Enrolment Rates Are Climbing. So What Explains the Sorry State of India’s Education Sector?


Another startling fact is that about 17% of students dropped out because they failed in their studies. Current government policy doesn’t allow schools to fail students until grade 8. As the ASER report points out, while the intention of the policy is commendable, there need to be measures in place to identify and focus on students who have fallen behind in the earlier grades. Currently, it would seem that the policy of not failing students has led to an adverse consequence where students left behind are not identified until they end up failing exams after grade 8.

Despite the fall in enrolment rates, over 86% of youth in the 14-18 range continue to be within the formal education system. Only about 5% are taking some type of vocational training, the majority of which are less than three months long.

Interestingly, a substantial proportion of youth in this age group is employed, irrespective of whether they are engaged in formal education or not. Overall, 42% of the youth is employed, including 39% of students engaged in formal education and 60% of students who have dropped out. Most of the work – over 70% – is on their own family’s farm. It is instructive to keep in mind here that ASER is a rural survey and urban deficiencies are still a black box when it comes to concrete data.

However, after accounting for work and enrolment in vocational courses, over one-third of the youth who have dropped out of education are not engaged in any kind of activity, i.e. neither studying, preparing for exams or employed – with nearly 75% of them being girls.

As shown in Figure 1, youth in the 14-18 age bracket didn’t have the skills expected of them from an elementary education and worryingly, there has been a declining trend in these skills. Apart from basic reading and arithmetic, ASER also conducted some testing on common everyday skills like counting money, reading maps, measuring length, calculating time, etc. The performance of the youth in these tasks was noticeably better.

The surveys use some specific metrics like access to mobile phones and bank accounts as proxies of youths’ exposure to the outside world. The findings are mixed: Approximately three-fourths of the youth surveyed had access to mobile phones and had their own bank account. But only about one-fourth had access to the internet and a computer, and only 16% had ever used an ATM.

In terms of aspirations, over 60% of youth in the age of 14-18 had aspirations to study beyond grade 12. Professional aspirations varied from military service and police for boys, to nursing and teaching for girls.

Policy interventions

After acknowledging the importance of early childhood education as well secondary education after age 14, the government is considering increasing the coverage of the RTE Act from 6-14 years to 3-16 years. However, as the current and previous ASER reports have demonstrated, merely increasing enrolment will not lead to the development of elementary skills that education is supposed to provide. If anything, the increase in enrolment rates over the last decade has coincided with the fall in the ability of students. Thus, stakeholders must take additional steps to ensure that the quality of education being imparted is not affected.

In that regard, two interesting policy interventions that are in the process of being rolled out hold some promise: same language subtitling (SLS) and outcomes fund for the education sector. Drop-out cases involving ‘loss of interest’ and ‘inadequate funds’ are problems that can be innovatively tackled. It is high time India develops its own indigenous Escuela Nueva model for the primary and secondary education space incorporating experiences from various cross-country models.

Conceived by Professor Brij Kothari of IIM Ahmedabad, SLS is the concept of subtitling existing Bollywood film songs on television. Kothari’s research estimates a 9% increase in the number of functional readers who watch TV programmes with SLS within a period of two years.

Similarly, a large outcome-based fund for education is all set to launch in India in early 2018. Touted as one of the first and largest funds for social enterprises, the fund would invest in education providers to work with government-run schools to deliver outcomes. There could be a variety of outcomes like early childhood interventions, retention of girl students, learning in primary schools and employability of students after high school. The fund is being launched by the Global Social Impact Investment Steering Group, an organisation comprising 13 member countries (including India), with a focus on channelling global social impact investment. Outcome fund based models are actively being employed by nations across the globe to fund social projects and have the potential to deliver the necessary outcomes.

While the statistics borne out by ASER 2017 do reveal a gamut of concerns regarding our secondary education sector, let’s take this as an opportunity to set up our house in order. While we may be gung-ho about increasing GDP growth rates or surpassing major economies by 2030, let’s also focus on translating these high numbers to meet the aspirations of India’s young and providing them with an education system that is innovative, proactive and prescient and yet deeply invests in foundational skills. The ‘dividend’ that we enjoy must not end up becoming a recipe for ‘disaster’.

Standard Deviation vs. Standard Error

Standard deviation and Standard Errors have been concepts that I have often erroneously mixed up and have struggled to differentiate between. As a result, I made a small presentation clarifying the difference between the two concepts.

To read the presentation, please click here

Enrolment Rates Are Climbing. So What Explains the Sorry State of India’s Education Sector?

[This article was published by TheWire.in on 21st December 2017. It was written in collaboration with Aniruddha Ghosh, a classmate from LSE. To read the article on TheWire.in, please click here . The article has also been posted on the LSE South Asia Blog and has been featured at Qrius]

Economic research shows that interventions aimed at improving cognitive skills rather than mere enrolment rates are required to boost economic growth.

Despite high enrolment ratios of over 96% in the last eight years, improvement in reading outcomes and arithmetic ability continues to be low. Credit: World Bank/Curt Carnemark/Flickr (CC BY-NC-ND 2.0)

Despite high enrolment ratios of over 96% in the last eight years, improvement in reading outcomes and arithmetic ability continues to be low. Credit: World Bank/Curt Carnemark/Flickr (CC BY-NC-ND 2.0)

Come January 2018, the Annual Status of Education Report (ASER) 2017 will be up for discussion among policy experts. Based on household-based surveys that cover children in the age group 3-16 across almost all rural districts of India, ASER provides estimates of children’s schooling status and their ability to do basic reading and arithmetic tasks.

ASER’s 2016 report observed that despite high enrolment ratios of over 96% in the last eight years, improvement in reading outcomes and arithmetic ability continues to be low. Moreover, a large proportion of students in both government and private schools continue to be below ‘grade level’. Grade level means that a student can deal with what is expected of her in that grade.

The ASER 2017 is targeted to look ‘beyond basics’: the age group between 14-18, primarily those outside the Right to Education ambit. There is a significant dearth of information in this regard and therefore, ASER 2017 will be a critical information asset to assess India’s madhyamik shiksha scenario. The government’s flagship Rashtriya Madhyamik Shiksha Abhiyan (RMSA) launched in 2009 and re-booted in 2013 as RMSA-Integrated has not been much of a success in India’s secondary education scene. In light of these observations, it is likely that ASER 2017 will throw up systematic issues that have continued to plague our secondary education.

ASER 2016 and other previous reports

Let’s take a step back and illustrate how our performance has been in the age group 3-16. Here is a snapshot of ASER’s previous reports on the state of India’s early grade development.

Figure 1: Percentage of children not enrolled in school. The trend for enrollment has been on the rise. Source: ASER 2016 Report

Figure 1: Percentage of children not enrolled in school. The trend for enrolment has been on the rise. Source: ASER 2016

Figure 2: Children in class III who are at ‘Grade Level’ 2008-2016. The lack of commensurate ability with grade level is evident. Source: ASER 2016

Figure 2: Children in class III who are at ‘Grade Level’ 2008-2016. The lack of commensurate ability with grade level is evident. Source: ASER 2016

Fig 3: Percentage of students who can at least do subtraction in Grade III. Large variation across states. In 2016, the number of children at grade level ranged from 50% in Himachal Pradesh to 10% in Uttar Pradesh. Source: ASER 2016 Report

Enrolment levels have been high for the 6-14 age group, and around 96% since 2009 onwards.

The percentage of children in grade 3 who are able to read at least a grade 1 level text has improved marginally – from 40% in 2014 to 42.5% in 2016.

Percentage of grade 3 children who could do two digit subtraction has marginally improved from 25% in 2014 to 27% in 2016. This has been the first year since 2010 when there has been an upward trend observed in arithmetic ability.

However, trends over time point to a dismal outlook. Figure 4 demonstrates the ability of grade 4 children in successive cohorts to read and do basic arithmetic. One can see a downward trend in the ability of successive cohorts.

Figure 4: The graphs show the performance of three cohorts from class IV to class VIII. The graph on the left shows the percentage of students who can do division; the graph on the right shows the percentage of students who can read a class II 2 level text. Source: ASER 2016

Figure 4: The graphs show the performance of three cohorts from class IV to class VIII. The graph on the left shows the percentage of students who can do division; the graph on the right shows the percentage of students who can read a class II 2 level text. Source: ASER 2016

Economics, enrolment rates and cognitive ability

One of the most influential studies to look at the relationship between education and economic growth was professor Lance Pritchett’s  Where has all the Education Gone? His 2001 study found no significant positive relationship between educational attainment and economic growth.

He wrote:

“In the decades since 1960, nearly all developing economies have already seen education attainment grow rapidly. The cross-national data show, however, that on average, education attainment contributed much less to growth than would have been expected in the standard augmented Solow model.”

He proposed several possible explanations for this phenomenon, the most important of which being the classic quantity-quality argument: Quality of educational attainment was so low that despite high educational attainment there wasn’t a significant increase in cognitive skills and human capital.

As outlined above, the very idea that improved schooling (through the lens of school attainment rates) – which has often been a cornerstone of most interventionist strategies – will raise economic well-being has often been discounted by economists. Eric Hanushek and Ludger Woessmann in their authoritative paper, ‘The Role of Cognitive Skills in Economic Development (2008)‘ find strong evidence that the cognitive skills of the population – rather than school attainment rates – are powerful determinants of earnings, economic growth and income distribution.

Their work premises itself on a rather simple question: whether education is the steering force or merely one of the several factors that are correlated with more fundamental development forces, say cognitive skills when it comes to economic growth. Hanushek-Woessmann armed with strong econometric evidence argue that cognitive skills overwhelmingly outnumbers schooling attainment rates when it comes to influencing economic growth. Moreover, the effect of cognitive skills on economic growth is larger in developing countries than in the developed ones.

 

Building on the works of Pritchett, and Hanushek and Woessmann, a recent research paper titled Does One Size Fit? The Impact of Cognitive Skills of Economic Growth by professor Nadir Altinok of the University of Lorraine and professor Abdurrahman Aydemir of Sabanci University studies the difference in the impact of cognitive skills on economic growth between developing and developed economies.

In addition to the findings mentioned above, the heterogeneous effects of cognitive skills vis-a-vis the income levels of the economies is rather stark: the magnitude of the effect of cognitive skills is about 60% higher for low-income countries compared to high-income countries, and this more than doubles when low total-factor productivity (TFP) countries are compared to high TFP countries. From a policy perspective, this encourages the view that the promotion of education policies that focus on the quality of education has especially larger payoffs in the least developed regions through the productivity channel. To sum it up, economic research overwhelmingly supports the idea of interventions aimed at improving cognitive skills rather than mere enrolment rates.

To cite a success story, Colombia has made impressive progress towards universal enrolment in basic education and at the same time has raised learning outcomes. A lot of this accrues to the flexible ‘new school’ model, commonly known as Escuela Nueva.

Escuela Nueva accepts multigrade teaching as an unavoidable condition in small schools of rural areas. It encourages to develop special materials and teaching methods for multigrade teaching. The academic achievement of the students in Escuela Nueva has been consistently higher than in urban schools. There are plenty of cross-country experiences to learn from and we should actively explore and replicate them here.

India’s scenario

Sarva Shiksha Abhiyan (SSA) was launched in 2000 to spread the availability of universal elementary education across India. Under SSA, commendable progress has been made in increasing enrolment rates; as well as providing basic infrastructures such as classrooms, water, toilets and boundary walls to all schools. Yet, what is the scenario when it comes to learning outcomes?

India made its debut in the Programme for International Student Aptitude (PISA) test in 2009 with 16,000 students from 400 schools across Himachal Pradesh and Tamil Nadu. While China – also a first-timer in 2009 – stormed into the number one position with Shanghai schools topping in math and science, India was at a paltry 72nd among the 74 participating countries. Since then, India has boycotted the PISA rankings citing ‘methodological differences’ but it plans to return to the rankings fold in 2021. Coupled with ASER’s findings over the years, this paints a sorry state of India’s primary education sector.

Now that enrolment rates are high, we would need to look for innovative interventions to improve learning-based outcomes in India. One of the recommendations of the ASER report, which the government has been focussing on in the last few years, has been early-childhood care and nurture, especially for children in the 0-3 age group. In fact, ASER’s study on three states (Andhra Pradesh, Assam and Rajasthan) found a positive and significant relationship between early childhood care and nurture and early grade learning outcomes. Although issues do remain in implementation, this avenue holds promise to further improve outcomes in India.

Additionally, in India, there are two interesting policy interventions that are in the process of being rolled out: same language subtitling (SLS) and outcomes fund for the education sector.

Conceived by Professor Brij Kothari of IIM Ahmedabad, SLS is the concept of subtitling existing Bollywood film songs on television. Kothari’s research estimates a 9% increase in the number of functional readers who watch TV programmes with SLS within a period of two years.

Similarly, a large outcome-based fund for education is all set to launch in India in early 2018. Touted as one of the first and largest funds for social enterprises, the fund would invest in education providers to work with government-run schools to deliver outcomes. There could be a variety of outcomes like early childhood interventions, retention of girl students, learning in primary schools and employability of students after high school. The fund is being launched by the Global Social Impact Investment Steering Group, an organisation comprising 13 member countries (including India), with a focus on channeling global social impact investment. Outcome fund based models are actively being employed by nations across the globe to fund social projects and has the potential to deliver the necessary outcomes.

While we wait for ASER’s 2017 findings, much of the theory and evidence that we have strongly suggests that raising enrolment rates hasn’t been enough to push our growth frontiers. The hope is that the required stakeholders will aim, plan and push for innovative interventions that encourage student achievements.

Finite Populations: How do we think about them?

Consider how a normal regression output on STATA looks.

Sample Regression Output

 

We find that along with the coefficients, we get other values like standard errors, p values, confidence intervals, etc. So what is the purpose of these stats?

The standard assumption in stat-econ literature is as follows: i)the observed sample is a random representative sample from the entire population, ii) we are interested in the true population parameters.

What does this mean? We assume that we pick a random representative sample from a population. After running a regression, we get a particular coefficient on all the x’s. We assume that in repeated sampling, we would get a distribution of coefficients for the x’s. In fact, one of the fundamental properties of a good estimator is unbiasedness or that the Expected Value of the coefficient from the sampling distribution is equal to the true value, i.e. the true coefficient is equal to the expected value of the coefficient in repeated sampling. Although we rarely do repeated sampling in practice, we do check for the significance of the coefficients with a null hypothesis that they are equal to zero. Thus, in a normal regression framework, the population has the true coefficients and is quite distinct from the samples which have a distribution of coefficients.

So what happens when we do have the entire population?

This is a tricky question, as we usually always assume that the population is infinite and we can never have the entire population. Thus, in cases of finite population, like the 28 states of India, it becomes a little bit tricky as to how one should interpret the regression output. I will attempt to summarise two popular views in the stat-econ literature in this regard.

Irrelevance

One view says that if you have the entire population, the p values and t statistics are irrelevant.  The finite population is considered to be a fixed set of elements. The coefficients derived are the true relationship since you have the entire population and not a sample of the population. Thus, the concepts of hypothesis testing and significance become meaningless for an entire population. This is because the stats and tests are only relevant for a sample, and not the entire population. The caveat here is that the model has to be correctly specified.

Super Population / Underlying Process

This view states that the observations in the population are simply a sample from an infinite population. For example, if we’re looking at the 28 states in India, we can interpret the set of observations at one time period as a sample; and the set of observations across all time as the infinite super population.

This becomes important, especially if you want to make inferences not just about relationships today but also if you want the relationships to hold for similar groups in the future.

Another related but slightly different view would be that the observed outcomes in the population are the products of an underlying process. In that regard, what standard errors and other test stats capture become relevant. According to Abadie et al (2014) “there are for each unit missing potential outcomes for the treatment levels the unit was not exposed to.”

On a similar vein Wallis and Robers (1956) claim, “the totality of numbers that would result from indefinitely many repetitions of the same process of selecting objects, measuring or classifying them, and recording results.”

In these regard, test stats become important in understanding whether the coefficients generated are true or have been generated as a result of a chance outcome.

 

References

  1. Abadie A, Athey S, Imbens G, Wooldridge J (2014), Finite Population and Causal Standard Errors, NBER Working Paper Series.
    Available at: http://www.nber.org/papers/w20325.pdf
  2. Asali M (2012), Can I make a regression model with the whole population? [Msg 1], ResearchGate.
    Message posted: https://www.researchgate.net/post/Can_I_make_a_regression_model_with_the_whole_population
  3. Frick R (1998), Interpreting statistical testing: Process propensity, not population and random sampling, Behavioral Research Methods, Instruments, & Computers. Vol 30 (3), pp: 527-535
  4. Hartley H, Sielker Jr R (1975), A “Super Population-Viewpoint” for finite population sampling, Biometrics, Vol. 31 (2) , pp: 411-422
  5. Hidiroglou, Michael Arsene (1974), Estimation of regression parameters for finite populations , Iowa State University Digital Repository.
    Available at: http://lib.dr.iastate.edu/rtd/5146
  6. January (2013), How to report data for an entire population? [Msg 1], CrossValidated.
    Message posted to: https://stats.stackexchange.com/questions/70296/how-to-report-data-for-an-entire-population

 

 

Simple Guide to Getting Orthogonalized Impulse Response Functions

Vector Auto-Regressions or VARs are used in time series analyses when there may be inter-dependencies or relationships among multiple time series.  For example, we may want to understand the relationship between GDP, current account balance, and inflation rates. Running a normal OLS regression would be inappropriate as each variable affects the other variables; OLS estimates would have an endogeneity problem or the estimates would be biased. In such scenarios, we use VAR methods.

Recently, I was working on some time series data that had the issues of reverse causality. As a result, I had to use a VAR model to get orthogonalized impulse response functions (OIRFs) in order to understand the relationship between variables. In the attached presentation, I describe the theory behind this orthogonalization as well as the steps to generate OIRFs on STATA.

To check the presentation, please click on this link

Older posts

© 2018 Sujan Bandyopadhyay

Theme by Anders NorenUp ↑