DATA MANAGEMENT

Datum
- singular, quantitative pieces of info

Data - plural

Data Set - number of series of data (GDP, UN, INT)

Observations - individual datum in a series. (inflation from 1970.01 to 1979.12 would be 120 individual observations)

Variable - measurable characteristic about which we can collect. Xs, Ys, etc. - consumption, income, # children, marital status, race, etc - MICRO

MACRO - GDP, inflation, ex-rates, etc.

Rectangular/Matrix Data Set. Variables across the top. Observations down the side. Spreadsheet.

i = 1 to n, where n is the total number of observations.

Xi, Yi, Zi. X1 to Xn. Y1 to Yn. Z1 to Zn.

Data collection - In contrast to other disciplines, economists don't usually collect original data. Most data is available from government agencies - Fed Reserve, Dept of Commerce, IMF, World Bank, Dept of Labor, OECD, Dept of Agriculture, Dept of Census, etc. Or available from industry groups - NYSE, NARealtors, National Mortgage Assoc, etc.

CROSS-SECTION DATA - Observations are data for different entities (persons, firms, states, countries) at the same point in time.

Examples:
a. survey data from interviews in response to questions about some variable of interest.
b. survey about state spending on education across the 50 states.
c. sales tax collection by county in Mich
d. per capita income in 50 countries.

Dummy variables - Qualitative variable. 0 or 1.
0 = white 1 = non-white
0 = F    1 = M
0 = union   1=non-union

Can be used for race, sex, marital status, etc. Also for event studies.

TIME SERIES DATA

Observations for the same variable over time - months, quarters, years, etc.

Example: Annual observations for 25 years, 56-80 for 12 variables. GNP, PCE, DPI, M1, etc.

Also: Cross sectional-time series.

Example: Xit, i=US, UK, Japan, GER, FR. t= 56-80.

CHANGE AND GROWTH -

Often we want to generate a variable from another variable.

1. Absolute Change = Delta Yt = Yt - Yt-1
Can be negative, we are NOT dealing with "abs value", but absolute change. Change in a variable from one period to another. Change in the budget, change in the deficit, example.

EVIEWS: genr dx = d(x)
genr dgdp = d(gdp)

Tells us how fast the variable is changing, measured in the units of the variable, example dollars.

2. Percentage Change (or Rate of change, Rate of growth, percentage growth)

% Change at time t = Change Y / Yt-1 = (Yt - Yt-1)/Yt-1 = (Yt/Yt-1) - 1
Multiply by 100 to scale up to a whole number.

EVIEWS: genr grgdp = d(gdp)/gdp(-1) * 100 (for annual data)

Tells us how fast the variable is growing, in relative terms, as a percentage.

Problems with percentage:

100 to 200 to 100

+100% - 50% net= +50%

Impose constant growth, or continuous compounding by using logs.

EVIEWS: genr inf = log(CPI/CPI(-1))*1200

100 to 200 to 100

+69.31%    -69.31%

CONSTRUCTED VARIABLES

Sometimes the desired data has to be constructed from "raw" data. Examples:

Growth rate of GDP, instead of GDP in dollars.

Four examples:

1. Index values - some data are presented in index form. Examples: PPI, CPI, IP, Index of Leading Economic Indicators, etc. Some base year is arbitrarily chosen, Index set at 100 in base year.

Growth rate is then calculated. Inflation = growth rate in the CPI. Percentage change from the period before or the percentage change from a year ago.

2. Nominal vs. Real. (Current vs constant) In many cases, we want to adjust nominal variables into real variables. Real GDP, real budget deficits, real Stock Returns, real income, real wages, etc. We need two variables - Nominal Variable and a Price Index - we can then calculate the variable in REAL TERMS.

Example: We have Nominal GDP (GDP) and GDP Deflator (GD) from 59-96. We want to construct Real GDP (RGDP):

REAL GDPt = NOM GDPt x  (100 /GDt)

where 100 is the GD in the base year.

Example:  YEAR          GDP(current \$)  GD
85                 4200                95
86                 4400                97
87                 4500               100
88                 4700               104
89                 4800               108

RGDP:           85 = 4200 X 100/95 = 4421 (constant 1987 \$)
86 = 4400 X 100/97 = 4536
87 = 4500 X 100/100 = 4500
88 = 4700 X 100/104 = 4519
89 = 4800 X 100/108 = 4444

3. Variables as a ratio of two others: DEF/GDP, FED/GDP, utilization rate (Food stamp example), percentage of school district getting free lunch, percentage of children passing exam, etc...

4. Converting all variables to logs. Log form.

Example: genr lgdp = log(gdp)

5. Converting to a different frequency. Monthly to quarterly or annually. Quarterly to monthly.

Example: NIPA accounts are quarterly. GDP, GNP, Deficits, GDP Deflator, Government expenditures, PCE, etc.

Most other statistics are released monthly: IP, CPI, PPI, interest rates, ex-rates, money supply, etc.

If you wanted to test for the effects of deficits on economic activity or interest rates, you would have variables with two different frequencies. You would have to convert everything to quarterly frequency, or annual.

EVIEWS: Converting Quarterly or Monthly Data to Annual

1. "store" the variable you want to convert (e.g. monthly or quarterly frequency)
2. start a new workfile, select annual frequency.
3. select "options" then "Dates-Frequency" to get conversion menu.
(Options: Average, Sum, First Obs, Last Observation).
4. "fetch" the M or Q variable and conversion will happen automatically.

6. Seasonal adjustment. SA/NSA. Randomness issue.

EVIEWS: for Q and monthly data only.
1. type: "seas x" to seasonally adjust variable x.
2. select the type of seasonal adjustment.