What's in an Economic Model?

Proud Economy 0002

Examining the Real Gross Domestic Product, 1 Decimal (GDPC1):

So if we look at the GDP from wikipedia, we get the breakdown. I suspose that if we created a spreadsheet we'd see that the total adds up to the nineteen industries and we could track them over time. Somehow, those aren't the nineteen that I would have guessed. What happened to “technology” or “real estate”?

Here's my version:

MEASURINGWORTH: United States Real GDP per capita (data):

y <- Quandl("MWORTH/0_5", transformation = "diff", type = "xts")
layout(matrix(c(1, 1, 2, 3), 2, 2, byrow = TRUE))
plot.ts(y, main = "MWORTH/0_5", ylab = "MWORTH/0_5")
acf(y, main = "Autocorrelations", ylim = c(-1, 1))
pacf(y, main = "Partial Autocorrelations", ylim = c(-1, 1))

Back to wikipedia:

GDP by industry

Industries by GDP value added 2011.

Industry	GDP value added $ billions 2011	% of total GDP
Real estate, renting, leasing	1,898	13%
State and Local Government	1,336	9%
Finance and insurance	1,159	8%
Health/social care	1,136	8%
Durable manufacturing	910	6%
Retail trade	905	6%
Wholesale trade	845	6%
Non-durable manufacturing	821	6%
Federal Government	658	5%
Information	646	4%
Arts, entertainment	591	4%
Construction	529	4%
Waste services	448	3%
Other services	447	3%
Utilities	297	2%
Mining	290	2%
Corporate management	284	2%
Education services	174	1%
Agriculture	173	1%
Total	15,075	100%

Examining the industries as interlinked, using each others products and services, leads to an input-output model. I wonder how much an average person could do with that.

.

ECON101 version of the Economy

If we look at the economy from this point of view, I suppose we have the agents: firms, consumers, government, and international. That leads to supply vs demand, investment vs savings, goods and money, microeconomics, macroeconomics, and international trade. As I recall in university classes many years ago, these were all static models. We didn't have computers available to be able to crunch the numbers for dynamic models then. I wonder what's been done since. In googling for dynamic models in economics, with multiple differential equations that may lead to chaos theory, there doesn't seem to be any analysis worked out there yet.

“The new science of chaos came about through weather analysis. Starting from the premise that economics is equally unpredictable, this original new book explores the ways that chaos theory may be used for economic analysis. The author shows that, since chaos theory sets out to demonstrate erratic and unpredictable behavior in a situation of total cause and effect, it has much to offer in understanding human society and the unpredictable nature of economics. It has always been assumed that the highly irregular behavior of economic time series was the consequence of extra-economic disturbances such as political decisions, trade unions, the weather, foreign trade, etc. Goodwin makes it clear that there are not one, but two explanations of this confusing behavior.” 1

By accumulating the actions of each class of agents we might be able to use the central limit theorem (CLT) and normal distristution theory to scale to a macro-model. We need to research how much of this is practical.

Economic Indicators

The Conference Board is currently (June 12, 2013) predicting 1.1% growth in GDP for the second quarter. Their U.S. Economic Forecast chart forcasts Real GDP, Real Consumer Spending, Housing Starts, Real Capital Spending, and Net Exports. Those seem like a good starting point for a multi-variable model.

The federal government Economics & Statistics Administration does economic indicators,

“The Economics and Statistics Administration (ESA) releases 12 monthly and quarterly Principal Federal Economic Indicators collected by its constituent bureaus: the U.S. Census Bureau and the Bureau of Economic Analysis (BEA). Businesses rely heavily upon these indicators to make decisions every day. In their public comments, the Secretary and ESA’s Under Secretary and Chief Economist put the indicators into a national and global economic context.” 2

The Bureau of Economic Analysis publishes Input-Output information.

The Federal Reserve Bank of New York has the Economic Indicators Calendar. If we built a structure of models, then we could use the calendar to compare and measure results over time.

I wonder about forecasting and tracking by normal people. Does anyone summarize this type of information and point to published models for us? I suppose we'll have to collect some wikipedia pages:

“You have two cows; you want chickens; you set out to find another farmer who has chickens and wants a cow”. 3

Infrastructure for posting about models

So far I'm using R software and RStudio with markdown and “Knit HTML”. I notice that the figures don't translate into RSS streams. Google is terminating their reader in a few days anyway and I haven't determined where I'm going then anyway.

Here's my quick version of the Unit Labor Costs:

ULCMFG Manufacturing Sector Unit Labor Cost (data):

y <- Quandl("FRED/ULCMFG", type = "xts")
layout(matrix(c(1, 1, 2, 3), 2, 2, byrow = TRUE))
plot.ts(y, main = "FRED/ULCMFG", ylab = "FRED/ULCMFG")
acf(y, main = "Autocorrelations", ylim = c(-1, 1))
pacf(y, main = "Partial Autocorrelations", ylim = c(-1, 1))

I was creating pdf files with sweave and LaTeX, but the RStudio link to TeX doesn't properly escape single quotes and it fails whenever the R output contains them. I hadn't explored generating html from that setup yet either. I still have to conquer the LaTeX typesetting of the math so that we can study, document, and discuss the actual models. I'm putting urls into the text as bookmarks, so I can find them easily in the future. I haven't checked what markup is allowed in comments yet. Math symbols don't seem to be part of ordinary communication. I'd eventually like to create some sort of a reference from this, if I can keep it up.

Gary Young