5 of 37

Entrepreneurship

“FoodTech is the future and it is very important in our business growth. We believe FoodTech will help us understand our consumers and their consumption habits better. With this knowledge, we will be able to create better products for our consumers. With advancements in machinery technology, we will be equipped to efficiently create and produce our products.” Janice Wong, 2006,

Founder, 2am : dessertbar

6 of 37

Entrepreneurship

“Just do it! Entrepreneurship may only fit a small percentage of the population. It is a high risk, high reward journey. Most people will fail, so the average return may be lower than the traditional path. But just do it if you have a new idea. The cost is relatively small when we are young.” Ly Vu Thinh, 2016, Founder, VATICO (https://vatico.vn/)

7 of 37

Economics: New careers – data scientist

Data science = statistics + interpretation
Advantages of economics

Theoretical foundation, not “black box” data mining
Understand data generation process

Randomized controlled trials
Quasi-natural experiments

8 of 37

Data science

“My background in economics provides me with a strong foundation in understanding market dynamics and consumer behavior, which enables me to generate valuable insights into our target audience's preferences and trends … combined with my technical expertise in data engineering and analytics, allows me to deliver data-driven solutions that drive informed business decisions and optimize our strategies.” Liu Wusheng, 2014, Lead Business Intelligence Analyst, Dyson Operations Ltd

9 of 37

Data science

“Other than technical training like coding, Economics has provided me with different insights and perspectives. The unique combination which allowed me to make sense and meaning of data led to me pursuing a career as data analyst.” Fiona Lee, 2019, Data Analyst, Immigration & Checkpoints Authority

10 of 37

Data insights: Applying economics

11 of 37

Case study 1:

Work from home (WFH)

Suppose economists want to measure the impact of WFH on economic growth or productivity. → Need data on extent of WFH

One option is to survey workers to understand the trend → but costly.

12 of 37

Use other data:

If employees WFH, use less transport to get to workplace → affect transit ridership
If employees stay at home, they would use more electricity at home (e.g. for air-conditioning) → affect residential electricity consumption

I.P.L Png. "What MRT travel numbers tell us about work from home trends." Straits Times p.B6 (28 October 2022)

How to measure extent of work from home?

Note the difference between Apartment and HDB 2-room electricity consumption

→ scope for WFH varies

13 of 37

16 March 2020: US President Trump asked people to stay at home to slow the spread of the virus

→ While those in the wealthiest and poorest areas were both moving less than usual, those in the highest-income locations had already cut their movement by nearly half (“average median date”)

Inequality:

Work amenities -- save on commuting time, accommodate family needs
Less risk of illness

https://www.nytimes.com/interactive/2020/04/03/us/coronavirus-stay-home-rich-poor.html

Data insight: Inequality in WFH in the United States

Comparing top 10% income earners (blue) and bottom 10% income earners (orange), of 25 largest metro areas in the United States:

Percent change in movement: movement for the day of the week / average for the same days of week in January and February, with the exception of holidays. The top 10 percent and bottom 10 percent of household incomes for each metropolitan area are based on median household income data from the U.S. Census Bureau, 2013-2017.

The data offers real-time evidence of a divide laid bare by the coronavirus pandemic — one in which wealthier people not only have more job security and benefits but also may be better able to avoid becoming sick.

"Methodology: Cuebiq (which tracks about 15 million cellphone users nationwide daily) calculated distance traveled by measuring a line between opposite corners of a box drawn around the locations observed for each person on each day. The travel for each census tract is the median of these per-person distances.

The Times calculated the percent change in movement by comparing the movement for the day of the week with the average for the same days of the week in January and February, with the exception of holidays.

The top 10 percent and bottom 10 percent of household incomes for each metropolitan area are based on median household income data from the U.S. Census Bureau, 2013-2017 American Community Survey 5-Year Estimates for census tracts.

14 of 37

https://www.nytimes.com/interactive/2020/04/03/us/coronavirus-stay-home-rich-poor.html

Given stay-at-home orders, in metro areas with greater disparity between the richest and poorest residents, people in higher-income neighborhoods halted movement sooner than people in low-income neighborhoods.

United States: WFH inequality

Increasing disparity between richest and poorest residents

15 of 37

I.P.L Png. "What MRT travel numbers tell us about work from home trends." Straits Times p.B6 (28 October 2022)

Consider time period of Circuit Breaker (CB), Apr to Aug 2022

By Aug 2022, electricity consumption had come down to levels similar to before (dotted line ······ ) → suggests most people had ended WFH.

On the other hand, transit ridership was still markedly lower than before the pandemic (dashed line ------ ).

Why?

Curious disparity between trends in ridership vis-à-vis electricity

CB start

Return to Phase 2

16 of 37

I.P.L Png. "What MRT travel numbers tell us about work from home trends." Straits Times p.B6 (28 October 2022)

Hybrid work?: three days at site + two days remote
Workers, having been forced to find alternative transport during Covid-19, found more preferable options (e.g. private transport) which they continued to use after.
Remote work from other locations (not home)
Tourism still depressed

Potential reasons…

CB start

Return to Phase 2

17 of 37

Case study 2:

Benchmarking

How to improve business efficiency through benchmarking

18 of 37

New businesses are the life blood of the economy

Famous 20^th century economist Joseph Schumpeter described economic growth as process of creative destruction
Old businesses give way to new, bringing new product concepts, new ways of production, new methods of management

19 of 37

Who start new businesses?

Important to understand who are the people that start new businesses
Many factors affect decision to start a business
One in particular stands out because subject to behavioral bias – self-efficacy

20 of 37

Economic efficiency

For economic efficiency, limited resources should be allocated to most cost-effective use
At business level, entrepreneur should devote time and energies to most rewarding activity

Over time, market competition will weed out less efficient businesses
But why wait for time to take its toll?
Better way -- benchmarking

21 of 37

Benchmarking

In benchmarking, collect sales and cost information from businesses in the industry
Report to each business their relative performance – how they compare with others
Entrepreneurs who perform poorly (and over-rate themselves) will quit for better opportunities
More customers for entrepreneurs who perform well

22 of 37

Research

Researchers Hou Yun and Ivan Png carried out an experiment among 194 vendors in Singapore hawker centres
They randomly administered benchmarking to 60% of vendors
Subsequently

13% of vendors who received benchmarking quit – all were lower performers
6% of control vendors (who did not get benchmarking) quit

23 of 37

Conclusion

In process of creative destruction, old businesses must make way for new
To increase economic efficiency, the inefficient should close
Benchmarking is an effective way to accelerate the exit of inefficient businesses
It has been experimentally validated by the field research of Hou and Png

25 of 37

Case study 1:

Economic development

How to compare the economic development of various economies, or economic growth of a single economy, over time? What if reported GDP is not reliable?

26 of 37

Key Problem and Motivation

Policymakers need data of their variables of interest to devise and evaluate policies.

Ideally, the data we have is what we want.

The gold standard is to directly measure or attain, but…

May be costly to collect and measure

Some variables are complex and difficult to measure (e.g. culture, institutions, GDP)

Provenance might be questionable (can we trust authorities - do they have incentive to distort?)

May be unavailable (e.g. historical data)

Another approach: Use available, relevant observational data �(i.e. good proxies)

What insights may these non-economic observational data reveal?

27 of 37

China: GDP growth, 2007

Source: National Bureau of Statistics of China

National growth: 11.6%

28 of 37

Can you suggest any?

We need…

A common measure across space and time…

… that is not easily manipulated…

… and accurately represents economic output

29 of 37

“GDP figures are 'man-made' and therefore unreliable… volume data, such as power and rail freight and even (bank) credit, are interesting because there is less incentive to massage them at the local level. But they reveal only part of the truth, not the entire truth”

- Le Keqiang, then Party Secretary of Liaoning Province, PRC, March 12, 2007

https://www.reuters.com/article/us-china-economy-wikileaks-idUSTRE6B527D20101206

Why distort GDP? Some potential responses: amount of foreign aid received; political advancement (e.g. in China), electoral advantage for democratic regimes

From Martinez (2021): Governments generally have an incentive to overstate GDP growth, irrespective of the political regime under which they operate. In democracies, canonical models of political accountability predict that observable measures of government performance guide the behavior of voters at the polls (Ashworth, 2012). In this regard, GDP growth stands out as a prominent indicator used to assess the state of the economy and the performance of the incumbent. In autocracies, salient economic statistics can facilitate coordination and may trigger mass

political action against the ruling regime if the outcome is unsatisfactory (Edmond, 2013; Hollyer et al., 2015). Low economic growth in an autocracy may also cause the incumbent government to lose the support from a key constituency, such as the military (Bueno de Mesquita et al., 2004). Existing evidence shows that economic conditions affect political mobilization and government turnover in both democracies and autocracies (Leigh, 2009; Burke and Leigh, 2010; Br¨uckner and Ciccone, 2011; Manacorda and Tesei, 2020).

The manipulation of GDP growth statistics is facilitated by the fact that these figures are typically produced by countries’ national statistical agencies, which are usually under the control of the central government.4 These agencies estimate GDP growth based on information from multiple sources, such as banks, public utilities, and surveys of households and firms. The government itself is the main source of information on public spending and investment, oftentimes with little third-party verification. Hence, the exaggeration of economic growth can take place at multiple stages in the production of the official figures.

30 of 37

One solution: nighttime light (NTL) data

Difficult and costly to fake (vis-à-vis reported GDP)

Correlated with economic output

Data available for all countries…

Data available over time

31 of 37

https://www.moneyandbanking.com/commentary/2018/9/16/illuminating-gdp

What can NTL data reveal about differences in economic output and development?

Differences across space

(North Korea vis-à-vis South Korea):

Differences across time

(1992 → 2008):

32 of 37

https://www.vox.com/2015/1/6/7496749/city-lights-space

What can NTL data reveal about differences in economic output and development? - Western vs Eastern Europe

Western European capitals are brighter than Central European capitals in general, even though the latter are generally more populous.

33 of 37

What can nighttime light data reveal about differences in economic development and output?

Africa

Noor, Abdisalan M., Victor A. Alegana, Peter W. Gething, Andrew J. Tatem, and Robert W. Snow. "Using remotely sensed night-time light as a proxy for poverty in Africa." Population Health Metrics 6, no. 1 (2008): 1-13.

34 of 37

West Berlin

East Berlin

https://www.vox.com/2015/1/6/7496749/city-lights-space; https://www.businessinsider.com/divide-between-west-east-berlin-from-space-today-2019-11

Still, what are some potential concerns with using NTL data?

While West Berlin is richer, East Berlin emits more intense light per capita (against predictions). Why?

West Berlin uses 40,000 gas lamps → difficult to detect from space
Also differences in fittings and shades (for environmental reasons)

Note also the difference in color:

East Berlin uses more older low-pressure and high-pressure sodium lamps → emits orange light
West Berlin uses more modern LED and fluorescents → emits white light

The nighttime lights in East Berlin give off an orange hue, largely from older low-pressure and high-pressure sodium lamps. West Berlin, by contrast, emits more white light, likely because there are more modern LEDs and fluorescents.
West Berlin has boomed with shopping and commercial districts, compared to East Berlin, and the light reflects the higher levels of activity.
East Berlin emits more intense light per capita at night, even though the area is poorer. Throughout Germany — East German cities that used to be behind the Iron Curtain are brighter than West German cities of similar size. Why?

Daniela Augenstine, who worked in Berlin's street furniture department, told The Guardian that the east uses sodium-vapor lamps, which are older and produce a yellower light, while the west uses fluorescent lamps, which producer a brighter, whiter light. The west preferred non-sodium lamps because they were cheaper, easier to maintain, and better for the environment.

According to transport spokesman Ehlert, the difference has less to do with the type of lights used than with the fittings and shades.

https://apnews.com/article/ap-top-news-international-news-europe-fall-of-the-berlin-wall-berlin-933d2736c8574617810b403a5e11bf64

35 of 37

Martinez, Luis R. "How much should we trust the dictator’s GDP growth estimates?." Journal of Political Economy 130, no. 10 (2022): 2731-2769.

There is a positive relationship between the growth rate of NTL (more objective measure) and reported GDP growth for democracies and autocracies…
… but average reported GDP growth is systematically higher in autocracies than in democracies.

Data insight: To what extent do different political regimes distort reported GDP?

36 of 37

Martinez, Luis R. "How much should we trust the dictator’s GDP growth estimates?." Journal of Political Economy 130, no. 10 (2022): 2731-2769.

Define NTL elasticity of GDP: Percentage change in reported GDP growth given a percentage change in NTL intensity.

Recall: Price elasticity of demand (PED): the percentage change in quantity demanded given a percentage change in price.

Results: The NTL elasticity of reported GDP is higher for autocracies than democracies (i.e. autocracies report higher GDP growth for the same growth in NTL).

Data insight: To what extent do different political regimes distort reported GDP?

37 of 37

Economics: Career pathways -- traditional

Industry

Grab, Amazon, Facebook, Capitaland, SMEs

Banks and financial institutions

DBS, UOB, OCBC, Citibank, Standard Chartered, JP MorganChase, HSBC, Barclays, Bank of America, UBS, GIC, Temasek

Consulting

Accenture, KPMG, Ernst & Young

Public service

Ministry of Finance, Ministry of Manpower, Ministry of Trade & Industry, Ministry of Transport
Competition Commission, Economic Development Board, CPF Board, ICA

1 of 37

2 of 37

3 of 37

4 of 37

5 of 37

6 of 37

7 of 37

8 of 37

9 of 37

10 of 37

11 of 37

12 of 37

13 of 37

14 of 37

15 of 37

16 of 37

17 of 37

18 of 37

19 of 37

20 of 37

21 of 37

22 of 37

23 of 37

24 of 37

25 of 37

26 of 37

27 of 37

28 of 37

29 of 37

30 of 37

31 of 37

32 of 37

33 of 37

34 of 37

35 of 37

36 of 37

37 of 37