Economic Data Lit iteracy Edi Tebaldi, Ph.D. Professor of - - PowerPoint PPT Presentation

▶
economic data lit iteracy
SMART_READER_LITE
LIVE PREVIEW

Economic Data Lit iteracy Edi Tebaldi, Ph.D. Professor of - - PowerPoint PPT Presentation

Sep 04, 2023 46 likes •417 views

Economic Data Lit iteracy Edi Tebaldi, Ph.D. Professor of Economics, Bryant University Hassenfeld Institute for Public Leadership Practical men, who believe themselves to be quite exempt from any intellectual influence, are usually the

slide-1
SLIDE 1

Economic Data Lit iteracy

Edi Tebaldi, Ph.D.

Professor of Economics, Bryant University Hassenfeld Institute for Public Leadership

slide-2
SLIDE 2
  • “Practical men, who believe themselves to be quite exempt from any

intellectual influence, are usually the slaves of some defunct economist.” (John Maynard Keynes)

  • “Practical men, who believe themselves to be quite exempt from

mistakes interpreting data are often wrong or are pawns of data analysts and economists.”

slide-3
SLIDE 3

Outline

  • Data analysis in the era of big data
  • Assessing economic conditions
  • Common mistakes in data analysis
  • The DO’s and DON’TS of data analysis
slide-4
SLIDE 4

Data analysis in in the era of big ig data

Hassenfeld Institute for Public Leadership

slide-5
SLIDE 5

Why bother about economic data literacy?

  • Meet public administration’s modernization goals;
  • Make the decision-making process data-driven;
  • Data are been produced in unfathomable quantity:
  • Can you make sense of it?
  • What good are the data if you cannot analyze the information collected?
  • Are you asking your data the right/wrong questions?
  • Data scientists now how to process data, but are they the ones who know

best what is the right question to be asked?

  • Should we start with data-first answers and then works backward to

find the questions that should have been asked?

slide-6
SLIDE 6

Data Literacy

  • Data Discovery and Collection
  • Evaluating & Ensuring Quality
  • f Data and Sources
  • Data Organization
  • Data Manipulation
  • Data Conversion
  • Metadata Creation and Use
  • Data Curation, Security, & Re-Use
  • Data Preservation

Data Collection

  • Data Tools
  • Basic Data Analysis
  • Data Interpretation
  • Data Visualization
  • Identifying Problems Using Data
  • Presenting Data (story)
  • Data-Driven decision Making
  • Evaluating decisions based on data

Data Analysis & Application Data Management

slide-7
SLIDE 7

Analyzing Data

Descriptive Stats Presentation of basic statistics Big Data Analysis Extremely large data sets are analyzed computationally to reveal patterns, trends, and associations related to human behavior and interactions; Machine Learning Computational algorithms that provides the ability to (hopefully) automatically learn from the data and improve from experience without being explicitly

  • programmed. Prediction is

a key output of machine learning algorithms. Data Mining & Regression The practice of examining data in order to generate new information

slide-8
SLIDE 8

Assessing Economic Conditions

Hassenfeld Institute for Public Leadership

slide-9
SLIDE 9

Assessing economic conditions

  • Leading economic indicators: often change prior to large economic adjustments,

thus can be used to predict future trends.

  • Lagging economic indicators: reflect the economy’s historical performance and

changes to these are only identifiable after an economic trend or pattern has already been established.

slide-10
SLIDE 10

Lagging Economic In Indicators

National & Local

  • Current Economic Conditions

Indicator: RIPEC/Bryant & FED

  • Gross Domestic Product (GDP)
  • US, RI
  • Unemployment Rate: US, RI
  • Income: US, RI
  • Wages: US, RI
  • Consumer Price Index (Inflation):
  • US, New England

Global

  • Exchange Rates: US
  • Balance of Trade: US
  • Libor (Interest Rate): UK
  • Japan Stock market - NIKKEI
  • EURO-Zone Stock Market: EURO
slide-11
SLIDE 11

Warning: Do not let unemployment statistics fool you

Alternative Measures of Unemployment

  • Discouraged workers: would like to have

a job but they have not looked for work in the past four weeks:

  • Counted as out of the labor force;
  • Could be counted as unemployed but

they are not.

  • Involuntary part-time workers: people

who would like to work full-time but cannot find a full-time job.

  • Counted as employed.

BLS: https://www.bls.gov/lau/stalt.htm FRED: U6UNEM6RI

Labor Force Participation

slide-12
SLIDE 12

Leading Economic In Indicators Key Data Source: : https:/ ://fred.stlouisfed.org/

  • FED Leading Economic Indicators
  • USSLIND, RISLIND,
  • State Dataset:

https://www.philadelphiafed.org/rese arch-and-data/regional- economy/indexes/leading/

  • RIPEC/Bryant Leading Economic

Indicator

  • RI Briefing
  • Stock Market
  • SP500, DJIA, NASDAQCOM, VIXCLS
  • Manufacturing Activity
  • IPMAN, MANEMP (local: RIMFG,

SMU44000003000000011, PROV244MFG)

  • Corporate Business Profits
  • A446RC1Q027SBEA
  • Inventory Levels
  • TOTBUSMPCIMSA
  • Retail Sales
  • RRSFS (Local: RIERET, RIWRET )
  • Housing Market
  • S&P/Case-Shiller (local: RISTHPI )
  • Consumer Sentiment
  • UMCSENT
slide-13
SLIDE 13

Economic theory based Leading Indicators

Natural Unemployment rate (NROU) The Yield Curve (T10Y2Y)

slide-14
SLIDE 14

Common Mis istakes in in Data Analysis

Hassenfeld Institute for Public Leadership

slide-15
SLIDE 15

Data consumers often interpret the data incorrectly

slide-16
SLIDE 16

Nominal versus Real

  • Example 1: Wage growth
  • Fred dataset: https://fred.stlouisfed.org
  • Wages: Average Hourly Earnings of Production and Nonsupervisory Employees:

Manufacturing

  • 𝑆𝑓𝑏𝑚 𝑋𝑏𝑕𝑓𝑡 =

𝑋𝑏𝑕𝑓𝑡 𝐷𝑄𝐽

∗ 100

  • CPI: Consumer Price Index for All Urban Consumers: All Items, Index 1982-1984=100
  • How to change the base year? Multiply real wages by: 𝑫𝑸𝑱𝒄𝒃𝒕𝒇

𝟐𝟏𝟏

  • where 𝐷𝑄𝐽𝑐𝑏𝑡𝑓 if the CPI of the base year
slide-17
SLIDE 17

Levels versus Change (Growth rate)

  • It is important to look at both the level and the growth rate of a

variable.

  • The decision whether to use one or the other in economic analysis depends
  • n the question a researcher wants to answer.
  • Use levels to compare things that are measured in the same scale or

are of similar in size

  • Cannot directly compare population in TX to that of RI;
  • Cannot directly compare GDP (US$) to Population (number of people);
  • Use Growth Rates to compare things that are measured using

different scales or to identify how fast or slow an indicator is changing

slide-18
SLIDE 18

Levels versus Change (G (Growth rate)

Example 2: Population

  • Fred dataset:

https://fred.stlouisfed.org

  • Plot
  • Pop TX: “Resident Population in Texas”
  • Pop RI: “Resident Population in Rhode

Island”

  • Change Axis from “number of

people” to:

  • Percentage change
  • Percentage change from year ago

Example 3: GDP

  • RI GDP
  • Choose “ right” indicator
  • Change Axis from “US$” to:
  • Percentage change
  • Percentage change from year ago
slide-19
SLIDE 19

Tre rends and and Corr rrelation

66 68 70 72 74 76 78 80 82 50 100 150 200 250 1980 1982 1984 1986 1988 1990 1992 1994 1996 1998 2000 2002 2004 2006 2008 2010 2012

Correlation= 0.945

B A

66 68 70 72 74 76 78 80 82 50 100 150 200 250 1980 1982 1984 1986 1988 1990 1992 1994 1996 1998 2000 2002 2004 2006 2008 2010 2012 Australian males' life expectancy US Export Index (A)

Correlation= 0.945

B A 0.0% 0.5% 1.0% 1.5% 2.0%

  • 20.0%
  • 15.0%
  • 10.0%
  • 5.0%

0.0% 5.0% 10.0% 15.0% 20.0% 25.0% 1980 1982 1984 1986 1988 1990 1992 1994 1996 1998 2000 2002 2004 2006 2008 2010 2012 Australian males' life expectancy - % Change US Export Index (A) - % Change

Correlation= -0.07

B A

slide-20
SLIDE 20

Trends and Correlation

  • Time Series Data might have TRENDS
  • May cause severe statistical problems
  • Trending variables may produce unreliable estimates
  • Spurious Correlations
  • Examples: http://tylervigen.com/spurious-correlations

What to do:

  • Do not let the data analyst fool you;
  • Demand figures/tables with “variation in” or “% change in”
slide-21
SLIDE 21

Time series: “% change” is the way to go

1,000 2,000 3,000 4,000 5,000 6,000 7,000 8,000 9,000 10,000 50 100 150 200 250 300 350 400 450 500 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 Suicide - Suffocation R&S Spending (billion)

Correlation=0.975

R&D Spending Sucide - Suffocation

  • 4.0%
  • 2.0%

0.0% 2.0% 4.0% 6.0% 8.0% 10.0% 12.0% 14.0% 0.0% 0.0% 0.0% 0.0% 0.0% 0.0% 0.0% 0.0% 0.0% 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 Suicide - Suffocation R&S Spending (billion)

Correlation=0.12

R&D Spending Sucide - Suffocation

slide-22
SLIDE 22

Correlation versus Causation

Should we propose legislation to foster ice- cream consumption among children as a tool to increase academic performance?

slide-23
SLIDE 23

Correlation versus Causation

  • Policy making depends on your understanding of the

difference between correlation versus causation;

  • Correlation can assist examining what has happened, but

finding the cause of something gives you the opportunity to change it.

  • Correlations may happen by accident, but causal

relationships are the outcome of structural economic forces;

slide-24
SLIDE 24

How to identify fy Causation?

Causation, most of the times, cannot be identified only through graphical analysis or work with 2 variables.

  • 1. Ask the right policy question
  • 2. Design an experiment to make sure

that biases are eliminated or deploy appropriate statistical methods

  • 3. Conduct graphical

and statistical analysis

slide-25
SLIDE 25

Example

  • Question: Does class size affect children’s test scores?
  • Experiment:
  • What is the ideal experiment that could be used to capture the causal effect
  • f interest?
  • To test if class size affect children’s test scores, we could randomly assign some students

to small and other to big classes and then compare their academic performance by the end of the academic year;

  • Can we run/execute the ideal experiment?
  • Yes, do it!
  • No. Regression analysis may be able to answer the question of interest
slide-26
SLIDE 26

No Selection Controls Selection Controls Note: Standard errors are shown in parentheses. The sample size is 14,238. (1) (2) (3) (4) (5) (6) Private School 0.212 0.152 0.139 0.034 0.031 0.037 (0.060) (0.057) (0.043) (0.062) (0.062) (0.039) Own SAT Score/100 0.051 0.024 0.036 0.009 (0.008) (0.006) (0.006) (0.006) Predicted log(Parental Income) 0.181 0.159 (0.026) (0.025) Female

  • 0.398
  • 0.396

(0.012) (0.014) Black

  • 0.003
  • 0.037

(0.031) (0.035) Hispanic 0.027 0.001 (0.052) (0.054) Asian 0.189 0.155 (0.035) (0.037) Other/Missing Race

  • 0.166
  • 0.189

(0.118) (0.117) High School Top 10 Percent 0.067 0.064 (0.020) (0.020) High School Rank Missing 0.003

  • 0.008

(0.025) (0.023) Athlete 0.107 0.092 (0.027) (0.024) 0.110 0.082 0.077 (0.024) (0.022) (0.012) Sent Two Application 0.071 0.062 0.058 (0.013) (0.011) (0.010) Sent Three Applications 0.093 0.079 0.066 (0.021) (0.019) (0.017) Sent Four or more Applications 0.139 0.127 0.098 (0.024) (0.023) (0.020) Average SAT Score of Schools Applied to/100

Example: does private college education pays off? Dependent variable: Average Earnings

ln Yi = α + βPi + ∑ γj GROUPji + δ1SA Ti + δ2 ln PIi + ei

Regression Analysis

Is There a Causal Private School Effect?

Selection control is the key! Ambition and opportunity matters.

slide-27
SLIDE 27

The DO’s and DON’TS of Data Analysis

Hassenfeld Institute for Public Leadership

slide-28
SLIDE 28

Takeaway

Increasing need to make data-driven policy decisions, but data by themselves are not a panacea!

slide-29
SLIDE 29

The DOs

Rigorous and structured efforts are needed to:

  • Properly analyze the data;
  • Identify causal relationships.
slide-30
SLIDE 30

The DOs

What are the limitations of the data analysis and the possible impacts on the results? Cross-examine your worldviews and biases Discuss your question of interest & data work with

  • thers

Do I need a team of experts to properly analyze the problem at hand?

slide-31
SLIDE 31

The DON’TS

Don’t let correlations fool you Don’t let time series fool you Don’t let practical men fool you

slide-32
SLIDE 32

Source: AssetWorks

slide-33
SLIDE 33

Obrigado Gracias

slide-34
SLIDE 34

What is the Causal relationship of interest?

  • Does class size affect children’s test scores?
  • Does on-the-job-training increase productive?
  • Does private college education increase lifetime earnings?
  • Does R&D tax credits boost economic growth?
slide-35
SLIDE 35

Example: Big data in Education

  • Customized and dynamic

learning programs

  • Reframing course material
  • Grading Systems
  • Career prediction
slide-36
SLIDE 36

AI & Machine Learning Applications

slide-37
SLIDE 37

It is happening right now!

Urban Traffic

  • Pittsburgh: SURTRAC -Scalable

Urban Traffic Control-

  • SURTRAC is used to manage

traffic flows through several intersections and uses AI to

  • ptimize the traffic systems

toward reduced travel times, reduced number of traffic stops, and reduced wait times.

  • City reduced travel time by 25%,

traffic stops by 30%, wait time by 40%, and overall emissions by 21% during the course of the pilot. Predicting Fire Risk in Buildings

  • Atlanta: Fire Rescue Department

(AFRD) developed a predictive analytics software aimed at identifying buildings that have a higher likelihood of fire incidents.

  • The software accurately

predicted 73% of fire incidents in the building.

Chatbots

  • North Carolina agency's IT help

desk found made up more than 80 percent of its tickets were related to password reset, which chatbots tool care of.

  • Free staff for more complex tasks.