SPEEDING-UP BUSINESS CYCLE OBSERVATION WITH REMOTE SENSING

SATELLITE OBSERVED BUSINESS CYCLE INDICATORS (SOBI) AS A TOOL

Strictly confidential

®

Draft -Study commissioned by FFG-ALR, Österreichische Forschungsförderungsgesellschaft mbH-ALR

Vienna, November 2008

1

SOBI – Satellite Observed Business Cycle Indicators

Table of Contents

1. INTRODUCTION

2. CONVENTIONAL BUSINESS CYCLE ANALYSIS: CONCEPTS, FLAWS AND A CASE STUDY

2.1. CONCEPTS AND FLAWS

2.2. A CASE STUDY

3. REMOTE SENSING : APPROACHES AND ACCOMPLISHMENTS

4. TWO APPROACHES TO BUSINESS CYCLE ANALYSIS

4.1. THE BC-INDICATOR APPROACH

4.1.1.OVERVIEW4.1.2.EXAMPLES

4.1.3.THE BCI-APPROACH AND SOBI

4.2. ECONOMETRIC MODELS

4.2.1. THE BASICS

4.2.2. DISAGGREGATED DYNAMIC STOCHASTIC GENERAL EQUILIBRIUM (DSGE) MODELS

4.2.3. ECONOMETRIC SUB-MODELS PRONE TO SOBI AFFILIATION

4.2.4. MACROECONOMETRIC MODELS OF BUSINESS CYCLES AND SOBI

5. LINKING REMOTE SENSING TO TERRESTRIAL OBSERVATION: PROCEDURAL STEPS

5.1. CONVERSION OF SPATIAL DATA INTO STATISTICAL DATA5.2. COMMERCIAL SOLUTIONS ON THE MARKET

6. IMPLEMENTING REMOTE SENSING: BUSINESS CYCLE ANALYSIS IN APPROPRIATE SECTORS

6.1. THE TASK

6.2. HOUSING

6.3. AGRICULTURE

2

6.4. TRANSPORT

6.5. PRODUCTION CAPACITY AND EMISSION

7. CONCLUSIONS AND OUTLOOK

REFERENCES

ANNEX:

AN OVERVIEW OF BUSINESS CYCLE INDICATORS

SOBI: Satellite Observed Business Cycle Indicators, interne Studie im Auftrag der Agentur für Luft- und Raumfahrt (ALR) der FFG, Österreich, Dezember 2008, 105 S. (2. Preis beim Internationalen Wettbewerb )

https://www.gmxattachments.net/de/cgi/g.fcgi/mail/print/attachment?mid=babgega.1290614749.6421.hfmvquhefs.76&uid=V0RKR6F2HmFhYDtYQnU1lgowPnQId0fp&frame=content

3

1 INTRODUCTION

Business cycles reflect the changing pace of economic activity. The analysis and empirical representation of business cycles has an exceptionally long tradition in economics, both as a theoretical and an empirical science. Accurate analysis and forecasting of business cycles are key elements in developing successful economic policy. Failure to anticipate and counteract too heavy downward amplitudes leads to loss of production and income and rising unemployment. Thus smoothing the business cycle is a primary challenge for policy-makers.

By their very nature, the length of business cycles ranges from a few months to several years. Intervening in a rapid and timely fashion to avoid overshooting is amongst the most difficult – some say impossible – tasks in monetary and fiscal policy. The ability to observe and respond to real-time events is obviously a crucial factor in ensuring successful intervention. However, conventional statistical observation suffers from serious time-lags, as events take place, are recorded, communicated and finally transformed into statistical business cycle indicators. Clearly then, shortening the lag between events, their observation and subsequent interventions would be a significant advantage for economic policy-makers. Much effort over many years has been put into trying to improve the timeliness of observation of business cycle indicators. The closer one comes to real-time observation, the greater the effectiveness of business cycle analysis methods, whether depicted conventionally with indicators or more mathematically with econometric models.

The challenge, then, is to observe indicators of complex business cycle phenomena in real-time, and to transform these into statistical time series, minus the impact and occasional distortion of human interference, to significantly increase the accuracy of economic observations. Finally, it must be very clear that in the present circumstances of a recession, hypersensitivity to economic indicators, regardless of how accurately these reflect real-time events, is having a tremendous impact on economic behavior. For example, publication of new figures on new housing construction, though these may incorporate significant time-lags, can send the stock market down.

The aim of this study is to explore the potential of remote sensing via satellites to improve the timeliness of economic observations. It is clear that remote sensing cannot substitute for conventional business cycle indicators. But its potential to enhance the quality of these indicators needs to be investigated. Satellite observation is already a well-established tool to track weather, agriculture, natural disasters such as floods or environmental change. It has not, yet, been used to observe business cycle related phenomena. Of course most aspects of business cycle indicators cannot be observed visually. However, remote sensing could have significant potential to track developments in some areas, for example, housing, agricultural production or transport. This study examines what such a procedure would entail. As an interdisciplinary study attempting to bridge the fields of satellite remote sensing and economics, it runs the risk of appearing superficial in both areas. However, it will have accomplished an important purpose if it triggers in-depth research by specialists in both fields.

4

The study adopts a two-pronged approach to the subject, examining remote sensing and satellite observation on the one side and economic approaches to business cycle analysis on the other, in the hope of bringing them together.

The concept can be represented by a simple chart as follows:

This chart will accompany the study throughout, the relevant chapter in the text indicated by the color of the modules on the chart.

5

2 CONVENTIONAL BUSINESS CYCLE ANALYSIS: CONCEPT, FLAWS AND A CASE STUDY

2.1. CONCEPT AND FLAWS

What seems to be a relatively simple task, i.e. assessing the business cycle with the help of statistical time-series, is in fact an extremely complicated endeavor.

The following questions are far from trivial:

What is a cycle (described by length, amplitudes, fluctuations around a trend, etc.)

Should the cycle be measured around a growth path, and mathematically in absolute terms, with changing growth rates, deviations from the trend, including differences between the observed and the potential output (capacity gap).

What are the statistical indicators leading, coinciding with and lagging behind the reference cycle.

How can the cyclical component be extracted from time-series encompassing trends, seasonal fluctuations and residuals.

From this it is clear that assessing the pattern and the pace of economic activity has long been a major challenge for economic analysis. The analytical approaches are manifold. The three major schools, which are basically complementary, take the following approaches:

via business cycle indicators (BCI)

via econometric models

via surveys of company and/or consumer sentiment.

The first two will be utilized in this study.

In order to fulfill these and other criteria BCI must pass the following tests1:

“Conformity - the series must conform well to the business cycle;

Consistent Timing - the series must exhibit a consistent timing pattern, over time, as a leading, coincident or lagging indicator;

Economic Significance - cyclical timing must be economically logical;

Statistical Adequacy - data must be collected and processed in a statistically reliable way;

1 The Conference Board, Business Cycle Indicators Handbook,

6

Smoothness - month-to-month movements must not be erratic; and

Currency - the series must be published on a reasonably prompt schedule.”

All these are rigid criteria. Applied strictly, few indicators qualify by these standards2. The Conference Board itself acknowledges: “No quarterly series qualifies for lack of currency, and many monthly series lack smoothness.”3

Beyond the problems of conforming to business cycle theory and statistical adequacy, “timeliness” remains a major issue for the indicator approach. Mention has already been made of the difficulty represented by the time-lag between real events and indicators. This is a major risk to developing successful fiscal policy, since interventions may be poorly timed. It is worth remembering here Milton Friedman’s assertion that economic instability is primarily the result of untimely public policy intervention. Thus shortening the period between real events and our recognition of them is of utmost importance. This has lead the OECD recently to try to significantly improve analyses. “Effective business cycle analysis, and indeed the monitoring of a country's economic performance from a policy perspective, requires access to timely high quality short-term economic statistics (STES). Consequently in recent years there has been a lot of pressure on national statistics organizations (NSOs) to better serve their users by improving the timeliness of release for their short-term economic indicators. In response to this demand, NSOs have focused on improving the efficiency and methodology of their statistical production processes. So this begs the question: where would one look to find comprehensive documentation on good practices used by NSOs to improve the timeliness of their short-term economic statistics? The answer is the STES Timeliness Framework, a structured collection of documentation on a range of good practices currently used by NSOs for improving timeliness, reducing costs or improving accuracy for short-term economic statistics. This resource is freely available in the form of an intuitive, user friendly website developed by the OECD Short-Term Economic Statistics Expert Group.”4

However, whatever improvements in accuracy and timeliness may be achieved, there remains the human factor. Filling-out statistical questionnaires is not a high priority for companies. In addition, figures are quite often simply not available because of incomplete book-keeping procedures. Thus there may be insurmountable obstacles to providing quarterly or even monthly data and accuracy may be compromised. For financial data the problem may be less acute – which may explain why econometric models of business cycles lately rely almost exclusively on financial variables. For the “real sector”, however, the time-lag problem persists. Any improvement in the timeliness of BCI can enhance the quality of business cycle observation and thus the effectiveness of fiscal policy. Which brings us to the aim of this study. The ability to observe business cycle phenomena directly and without human interference would resolve the difficulties inherent in the current time-lag affecting data. Can remote sensing via satellite observation offer a new means to analyse events in real-time? Finding an answer to that question is an exhilirating, but not simple, quest.

2 A list of BCI is presented in the ANNEX3 The Conference Board, loc.cit.4 Journal of Business Cycle Measurement and Analysis 2007, vol.2007, no.1, pp.103-125; Report: Improving Timeliness

for Short-Term Economic Statistics; cf. www.oecd.org/std/research/timeliness

7

2.2. A CASE STUDY

At the risk of anticipating the contents of this study, the following example effectively demonstrates the fundamental “philosophy” of applying remote sensing to conventional business cycle analysis. The concept: SOBI should enable a significant reduction in the time-lag between actual events and our apprehension of those events. This would ensure that information about real events in the business cycle reaches policy-makers and the public much more quickly. That would allow more timely reaction to those events, and intervention to counteract undesirable trends and alleviate their effects. That is the scenario this case study exemplifies:

The potential impact of SOBI in a nutshell – Housing as a case study:

It is very well known that the housing sector was observed with scrupulous attention during the second half of 2008, at the latest, as the world financial crisis gathered momentum. Information on data pertaining to any aspect of the housing sector whatsoever had an immediate impact on stock and credit markets and, in turn, influenced institutional and political behavior. Release of data by the US Census Bureau was awaited with great impatience. In general, housing sector indicators feature frequently within the list of leading business cycle indicators5, and their date of release is eagerly anticipated:

US Census Bureau: 2008 Economic Indicator Release Schedule: by Date

Select: By Date - By Indicator |2008 At-a-Glance Version  

Indicator Release Date Time Period Covered

Construction Put in Place Friday, August 1, 2008 10:00am June 2008

Manufacturer's Shipments, Inventories, and Orders Monday, August 4, 2008 10:00am June 2008

Wholesale Trade Friday, August 8, 2008 10:00am June 2008

U.S. International Trade in Goods and Services Tuesday, August 12, 2008 8:30am June 2008

Advance Monthly Sales for Retail and Food Services

Wednesday, August 13, 2008 8:30am July 2008

Manufacturing and Trade: Inventories and Sales Wednesday, August 13, 2008 10:00am June 2008

Housing Starts and Building Permits Tuesday, August 19, 2008 8:30am July 2008

New Residential Sales Tuesday, August 26, 2008 10:00am July 2008

5 http://www.census.gov/epcd/econ/www/indijun.htm

8

Advance Report on Durable Goods Manufactures' Shipments and Orders

Wednesday, August 27, 2008 8:30am July 2008

Construction Put in Place Tuesday, September 2, 2008 10:00am July 2008

Manufacturer's Shipments, Inventories, and Orders Wednesday, September 3, 2008 10:00am July 2008

Quarterly Financial Report -- Manufacturing, Mining, and Wholesale Trade

Monday, September 8, 2008 10:00am 2nd Quarter 2008

Wholesale Trade Tuesday, September 9, 2008 10:00am July 2008

Quarterly Services Survey Thursday, September 11, 2008 10:00am 2nd Quarter 2008

U.S. International Trade in Goods and Services Thursday, September 11, 2008 8:30am July 2008

Advance Monthly Sales for Retail and Food Services

Friday, September 12, 2008 8:30am August 2008

Manufacturing and Trade: Inventories and Sales Friday, September 12, 2008 10:00am July 2008

Housing Starts and Building Permits Wednesday, September 17, 2008 8:30am August 2008

Advance Report on Durable Goods Manufactures' Shipments and Orders

Thursday, September 25, 2008 8:30am August 2008

New Residential Sales Thursday, September 25, 2008 10:00am August 2008

Construction Put in Place Wednesday, October 1, 2008 10:00am August 2008

Manufacturer's Shipments, Inventories, and Orders Thursday, October 2, 2008 10:00am August 2008

Quarterly Financial Report -- Retail Trade Wednesday, October 8, 2008 10:00am 2nd Quarter 2008

Wholesale Trade Thursday, October 9, 2008 10:00am August 2008

U.S. International Trade in Goods and Services Friday, October 10, 2008 8:30am August 2008

Advance Monthly Sales for Retail and Food Services

Wednesday, October 15, 2008 8:30am September 2008

Manufacturing and Trade: Inventories and Sales Wednesday, October 15, 2008 10:00am August 2008

Housing Starts and Building Permits Friday, October 17, 2008 8:30am September 2008

New Residential Sales Monday, October 27, 2008 10:00am September 2008

Housing Vacancies Tuesday, October 28, 2008 10:00am 3rd Quarter 2008

Advance Report on Durable Goods Manufactures' Shipments and Orders

Wednesday, October 29, 2008 8:30am September 2008

Construction Put in Place Monday, November 3, 2008 10:00am September 2008

Manufacturer's Shipments, Inventories, and Orders Tuesday, November 4, 2008 10:00am September 2008

Wholesale Trade Friday, November 7, 2008 10:00am September 2008

U.S. International Trade in Goods and Services Thursday, November 13, 2008 8:30am September 2008

Advance Monthly Sales for Retail and Food Services

Friday, November 14, 2008 8:30am October 2008

Manufacturing and Trade: Inventories and Sales Friday, November 14, 2008 10:00am September 2008

Housing Starts and Building Permits Wednesday, November 19, 2008 8:30am October 2008

Advance Report on Durable Goods Manufactures' Shipments and Orders

Wednesday, November 26, 2008 8:30am October 2008

9

New Residential Sales Wednesday, November 26, 2008 10:00am October 2008

Construction Put in Place Monday, December 1, 2008 10:00am October 2008

Manufacturer's Shipments, Inventories, and Orders Thursday, December 4, 2008 10:00am October 2008

Quarterly Financial Report -- Manufacturing, Mining, and Wholesale Trade

Monday, December 8, 2008 10:00am 3rd Quarter 2008

Wholesale Trade Wednesday, December 10, 2008 10:00am October 2008

Quarterly Services Survey Thursday, December 11, 2008 10:00am 3rd Quarter 2008

U.S. International Trade in Goods and Services Thursday, December 11, 2008 8:30am October 2008

Advance Monthly Sales for Retail and Food Services

Friday, December 12, 2008 8:30am November 2008

Manufacturing and Trade: Inventories and Sales Friday, December 12, 2008 10:00am October 2008

Housing Starts and Building Permits Tuesday, December 16, 2008 8:30am November 2008

New Residential Sales Tuesday, December 23, 2008 10:00am November 2008

Advance Report on Durable Goods Manufactures' Shipments and Orders

Wednesday, December 24, 2008 8:30am November 2008

Immediately after the release of data on housing on October 17th, 2008, the stock market plunged on the shockwave of lower than expected results. A selection of citations found in the media indicates the impact:

1. BBC NEWS | Business | US stocks slide on housing data

 Page last updated at 21:06 GMT, Friday, 17 October 2008 22:06 UK ... US stocks slide on housing data ... MARKET DATA - 07:26 UK ...news.bbc.co.uk/2/hi/business/7675279.stm –

2. RealClearMarkets - AP - Markets - Oct 17, 2008 - US stocks open ...  17 Oct 2008 ... Videos. PRINT ARTICLE. |. SEND TO A FRIEND. October 17, 2008. US stocks open lower after housing data. Tim Paradis ...www.realclearmarkets.com/news/ap/finance_business/2008/Oct/17/us_stocks_open_lower_after_housing_data.html - 12k -

3. STOCKS NEWS US-Home improvement stores drop on housing data - News ... 17 Oct 2008 ... STOCKS NEWS US-Home improvement stores drop on housing data. | 17 Oct 2008 | 09: 57 AM ET. Text Size. NEW YORK (Thomson Financial) - Stocks .www.cnbc.com/id/27237631 - 81k

4. European Union BreakingNews - information on BreakingNews in EU U.S. stocks slide on housing data. Friday 17 October 2008 15:40:10 - Breaking News - Source: Reuters Reuters. By Herbert Lash ...www.euractiv.com/en/BreakingNews?GUID=TRE49G4XJ&_xsl=Article - 34k -

5. 17 Oct 2008 ... US stocks open lower after housing data, Print · E-mail ... Last 6 Days - Business. Sorted by popularity Thursday, 16th of October 2008 ...www.heraldextra.com/content/view/284959/18/ - 63k –

6. Washington Times - US stocks open lower after housing data 17 Oct 2008 ... US stocks open lower after housing data. TIM PARADIS ASSOCIATED PRESS Originally published 10:08 a.m., October 17, 2008, updated 09:47 a.m., ...www.washingtontimes.com/news/2008/oct/17/us-stocks-open-lower-after-housing-data/ - 103k -

7. Khaleej Times Online - US housing data weak; US, France to meet  US housing data weak; US, France to meet. (Reuters) 17 October 2008 ... U.S. housing starts fell sharply in September and the world's biggest oilfield ...www.khaleejtimes.com/DisplayArticleNew.asp?col=&section=business&xfile=data/.../2008/October/...October646.x... - 30k

8. US ECONOMIC INDICATORS: Housing Data, Latest 6 Months  Fri, Oct 17 2008, 13:49 GMT http://www.djnewswires.com/eu. US ECONOMIC INDICATORS: Housing Data, Latest 6 Months Data seasonally adjusted except actual , ...www.fxstreet.com/news/forex-news/article.aspx?StoryId=f7322c24-064a-405d-a4f4-91a60bf1b423 - 45k -

10

9. U.S. stocks slide on housing data - swissinfo  -17 Oct 2008 ... 17.10.2008 - 15:40 U.S. stocks slide on housing data; 17.10.2008 - 15:03 ... October 17, 2008 - 12:01 PM. Growth forecast revised downwards ...www.swissinfo.ch/eng/news/international/U_S_stocks_slide_on_housing_data.html?siteSect=143&sid...ti - 43k

10. U.S. stocks slide on housing data | World | Reuters  17 Oct 2008 ... NEW YORK (Reuters) - A new batch of weak US housing data on Friday added ... U.S. stocks slide on housing data. Fri Oct 17, 2008 4:40pm BST ...uk.reuters.com/article/worldNews/idUKTRE49G4XJ20081017 - 62k - Im   Cache - Ähnliche Seiten

11. Bloomberg.com: Worldwide  U.S. Stocks Drop as Housing, Consumer Data Offset Buffett Buys. By Eric Martin. Enlarge Image/Details. Oct. 17 .... Last Updated: October 17, 2008 17:26 EDT ...

12. FT.com - Housing data weighs on Wall Street  17 Oct 2008 ... Mon Oct 13 00:00:00 EDT 2008 - 7 page pdf. Price: 25.0 ... US MARKETS (more) · Housing data weighs on Wall Street · Housing data weighs on ...us.ft.com/ftgateway/superpage.ft?news_id=fto101720081038206913 - 32k - Im   Cache

13. STOCKS NEWS US-Home improvement stores drop on housing data ...  Macroeconomic News. Friday, 17th October 2008. STOCKS NEWS US-Home improvement stores drop on housing data. 17-OCT-2008 14:57 ...www.lse.co.uk/MacroEconomicNews.asp?ArticleCode=6idn9ljz33rf2y4&ArticleHeadline...us...... - 34k –

The original press release on which these media reports are based reads as follows:

11

The relevant point for the present study is that these data released on October 17th reflect the available information on the previous month, September, and that this information is also only provisional, being subject to revision as stated in the explanatory notes included above. It is surely no exaggeration to say that any reduction in the time-lag between the actual situation and reporting of it and any improvement in accuracy of data (e.g. by reducing sampling errors and

12

biases) would be of considerable value in modulating the effect on stock markets and accelerating the formulation of appropriate policy responses. Remote sensing to gain information in the form of Satellite Observed Housing Indicators should contribute to a notable improvement in the available data.

13

3 APPROACHES AND ACCOMPLISHMENTS OF REMOTE SENSING

SatelliteSensor

Images Conversion ofSpatial Data

S O B I

(plus GIS)

Sub-sectors or

AppropriatSectors:

e.g.-Housing

-Agriculture-Transport

-……..

BC-IndicatorApproach

EconometricBC-Models

BC Sta-tis-tics

- Reduces observation-lagand

leads to timeliness- No bias through human interference

Remote sensing is not an unambiguous term. For the purposes of this study the following definition is relevant: “Remote sensing is the art and science of making measurements of the earth from sensors, such as cameras carried on airplanes, satellites, or other devices. These sensors collect data in the form of images. Remote sensing systems provide specialized capabilities for manipulating, analyzing, and visualizing images. Without strong geographic data management and analytical operations they cannot be called true GIS.” (Geographic Information Systems)6 As regards BCI, images taken from satellites rather than from airplanes appear to be the most significant. That is why this study has selected the term Satellite Observed Business Cycle Indicators (SOBI).

It may not be immediately evident to economists that satellite observation can be relevant to BCI. In this regard, it should be noted that remote sensing images have been used very successfully over recent decades, in particular in areas like the geosciences, response to natural disasters, weather, environmental protection, agriculture, urban sciences, archeology, intelligence etc. Little attention has thus far been paid to their relevance to economics. Undoubtedly, economic phenomena are more difficult to observe from space than the areas just mentioned. However, a review of some of the accomplishments in these areas suggests great promise in applying these methods to economics.

Firstly, missions conducted within the EOS (Earth Observing System) framework are a source of inspiration. This program of NASA , launched in 1997, comprises artificial satellite missions in earth orbit designed for long-term global observations of the land surface, biosphere, atmosphere, and oceans and is the core of NASA's Earth Science Enterprise (ESE):

6 Source: Simon Fraser University Library

14

Similar tasks are being undertaken in Europe by ESA. Earth observation with Envisat includes, for instance, missions providing scientific data on the chemical composition of the atmosphere, and monitoring oceans, vegetation, fires and environmental pollution7.

More specifically, applications are as follows:

Atmosphere

ASARGOMOS

RA-2MERISMIPASMWRLR 

SCIA

7 “Data from Earth Observation satellites are one of the most important assets brought to us by the space age. By helping us to understand our planet and secure our environment they benefit our daily lives in many ways. The future looks even more promising as new ways of using these invaluable data are discovered. The objective and continuous views of our planet supplied by satellite images and data provide scientists and decision makers with the information they need to understand and protect our environment. Among their many applications are monitoring the air, seas and land; providing the basis for accurate weather reports; and supplying national and international relief agencies with data when disasters strike.” Source: ESA

15

AATSRDORIS

Clouds

Water Vapour

Radiation Budget

Temperature / Pressure

Trace Gases

Aerosols

Turbulence

Land

ASARGOMOS

RA-2MERISMIPASMWRLR 

SCIAAATSRDORIS

Surface TemperatureVegetation Characteristics

Agriculture and Forestry

Surface Elevation

Geology and TopographyHydrology ParametersFloodingFire 

16

Instrument contributions to Envisat's mission objectives x   principal contribution (x)  secondary or experimental contributionSource: ESA, Envisat

Another example of a more specific application from the US can be found in the area of security. With the aid of a flow chart, Homeland Security describes the procedure by which decision-making is greatly assisted through satellite observation:

Beyond the examples mentioned thus far, it is useful for this study to take a closer look at how remote sensing works and is implemented.The overarching features of remote sensing can be illustrated by quoting MODIS.To explain, first, what MODIS is8:

“The MODIS Rapid Response System was developed to provide daily satellite images of the Earth's landmasses in near real time. True-color, photo-like imagery and false-color imagery are available within a few hours of being collected, making the system a valuable resource for organizations like the U.S. Forest Service and the international fire monitoring community, who use the images to track fires; the United States Department of Agriculture Foreign Agricultural Service, who monitors crops and growing conditions; and the United States Environmental Protection Agency and the United States Air Force Weather Agency, who track dust and ash in the atmosphere. The science community also uses the system in projects like the Aerosol Robotic Network (AERONET), which studies particles like smoke, pollution, or dust in the atmosphere. More information about science and application partners, including links, is provided on our applications page. Captioned interpreted images for educators, the media, and the public

8 http://modis.gsfc.nasa.gov/

17

are available through the Earth Observatory. The system is freely available to everyone--scientists, operational users, educators, and the general public.

The Moderate Resolution Imaging Spectroradiometer (MODIS) flies onboard NASA's Aqua and Terra satellites as part of the NASA-centered international Earth Observing System. Both satellites orbit the Earth from pole to pole, seeing most of the globe every day. Onboard Terra, MODIS sees the Earth during the morning, while Aqua MODIS orbits the Earth in the afternoon.”

Some most important areas of MODIS are the following:

Fires Floods Land Oceans Outreach Air Quality SensorWeb Ice Direct Broadcast

In order to underline the extraordinary potential of this kind of remote sensing in the field of SOBI, a few quotes may be given9:

“Fire managers can access daily maps showing the locations of active fires in the conterminous United States and Alaska. The United States Forest Service's Remote Sensing Applications Center produces daily regional fire maps based on MODIS Rapid Response Data.”

The MODIS Rapid Response System has been instrumental in developing a fire alert system used by Eskom, the South African power company, and Conservation International. Fire was badly affecting the South African power company’s ability to operate. In an average year, more fires burn in sub-Saharan Africa than any other part of the world, and some of these fires caused power outages. Fires burning near transmission lines can cause outages when flames ionize the air around the lines, breaking down the normal isolating properties of air. These conditions cause power arcs or surges on the lines, and the transmission protection system shuts down momentarily and recloses the breakers to restore the power.

The Flood Observatory uses MODIS to observe known floods around the world, and, on a limited basis, detect new floods. The maps they generate are available to governments and relief organizations. For flood detection, the Flood Observatory has defined at least 710 sections of rivers worldwide. Each section is 20 to 30 kilometers in length and 20 to 30 kilometers in width. MODIS monitors the sections, called reaches, to determine the status of each. When water levels rise to flood status, the river can be targeted by higher-resolution sensors.

The Dartmouth Flood Observatory also compiles yearly catalogs, maps, and images of river floods from 1985 to the present, primarily for researchers. The Observatory publishes a comprehensive Atlas of Global Flood Hazard online that illustrates the complete history of river flooding as observed by MODIS since launch. The Atlas is being used by the international insurance industry and by national and local government for assistance in determining flood hazard. 9 http://rapidfire.sci.gsfc.nasa.gov/apps/

18

The U.S. Department of Agriculture's Foreign Agricultural Service (FAS) is using MODIS satellite data to estimate the predicted yield of crops. MODIS provides daily, high-quality, photo-like images that can be used to observe large areas across the world. These images help FAS improve the accuracy and timeliness of the crop yield predictions, which are needed to make decisions affecting U.S. agriculture, trade policy, and food aid. MODIS products allow FAS analysts to distinguish between different crops like wheat and rice and permit analysts to measure other features like surface temperature and snow cover. Analysts can gauge the overall health of agriculture by comparing current data with previous years. MODIS products, first used by FAS in the summer of 2003, demonstrated their utility by helping analysts identify new areas of irrigated agriculture in the Middle East.

The MODIS Rapid Response System also provides data to the Aerosol Robotic Network (AERONET) program. The program is a collaboration of agency, institute, and university partners designed to monitor aerosols, liquid or solid particles suspended in the atmosphere. AERONET is a worldwide network of ground based sensors that monitor the air for aerosols to understand the impact of aerosols on climate change. The MODIS Rapid Response system provides daily near real time imagery of the sensor sites here. By comparing satellite and ground-based observations, scientists can learn how aerosols reflect and absorb light and can validate satellite-based aerosol observations. This will improve the tools scientists use to monitor aerosols over the entire Earth.”

Similarly, EOSDIS should be cited10:

The Earth Observing System Data and Information System (EOSDIS) manages and distributes data products through the Distributed Active Archive Centers (DAACs). The centers process, archive, document, and distribute data from NASA’s past and current research satellites and field programs. Each center serves one or more specific Earth science disciplines and provides data products, data information, services, and tools unique to its particular science.

Here also applications are widespread:

“Sensing Our Planet: NASA Earth Science Research Features

The 2007 volume highlights multidisciplinary research that uses Earth-observing data from NASA Earth science data centers. Articles explore research on land use, invasive species, groundwater availability, air pollution, and more, for readers with a range of scientific understanding.”

As examples, the 2007 reports and parameters using different types of data like ASTER, SAR, MODIS, LASE etc. include11:

10 http://nasadaacs.eos.nasa.gov/index.html11 http://nasadaacs.eos.nasa.gov/articles/2007_menu.html

19

Burgundy through space and time Land use

Can Earth's plants keep up with us? Biomass, population Gridded Population of the World

Connecting rainfall and landslides Rainfall, landslides

Getting at groundwater with gravity Gravity, groundwater

Saharan dust versus Atlantic hurricanes Aerosols, water vapor

After the Larsen B Land ice, elevation

Pinpointing an invasive plant's next move Land cover type

Following the World Trade Center plume Aerosols

Pollution trials for the Beijing Olympics Nitric oxide

Grasping the subtle needs of vegetation Vegetation indices Flux towers”

There are a vast number of articles on remote sensing applications, and those wishing to pursue the topic further could consult the following link: Geoscience and Remote Sensing Symposium, 2007. IGARSS 2007. IEEE International, where proceedings from earlier years are also available.

For the sake of illustration, where it may be of particular interest to economists, a few titles of articles on in-depth research already being carried out are given below:

Date: 23-28 July 2007High resolution COSMO - SkyMed SAR images for oil spills automatic detectionTrivero, P.; Biamino, W.; Nirchio, F.Page(s): 2-5Digital Object Identifier 10.1109/IGARSS.2007.4422715

Seasonal and interannual patterns of chlorophyll bloom timing in the Gulf of Cádiz:Navarro, G.; Prieto, L.; Huertas, I.E.; Ruiz, J.; Gomez-Enri, J.Page(s): 50-53Digital Object Identifier 10.1109/IGARSS.2007.4422727

Field measurement of Gobi surface emissivity using CE312 and Infragold Board at Dunhuang calibration site of ChinaYong Zhang; Zhiguo Rong; Xiuqing Hu; Jingjing Liu; Lijun Zhang; Yuan Li; Xingying ZhangPage(s): 358-360Digital Object Identifier 10.1109/IGARSS.2007.4422804

High resolution urban feature extraction for global population mapping using high performance computingVijayaraj, V. Bright, E.A. Bhaduri, B.L. Oak Ridge Nat. Lab., Oak Ridge;

20

Seasonal and interannual patterns of chlorophyll bloom timing in the Gulf of Cádiz:Navarro, G.; Prieto, L.; Huertas, I.E.; Ruiz, J.; Gomez-Enri, J.Page(s): 50-53Digital Object Identifier 10.1109/IGARSS.2007.4422727

Feasibility of spaceborne bistatic radar missions for land applicationsPietra, G.D.; Capobianco, F.; Falzini, S.; Pierdicca, N.; De Titta, L.Page(s): 93-96Digital Object Identifier 10.1109/IGARSS.2007.4422738

Field measurement of Gobi surface emissivity using CE312 and Infragold Board at Dunhuang calibration site of ChinaYong Zhang; Zhiguo Rong; Xiuqing Hu; Jingjing Liu; Lijun Zhang; Yuan Li; Xingying ZhangPage(s): 358-360Digital Object Identifier 10.1109/IGARSS.2007.4422804

Disaster monitoring and environmental alert in Taiwan by repeat-pass spaceborne SARChih-Tien Wang; Kun-Shen Chen; Hong-Wei Lee; Jong-Sen Lee; Boerner, W.-M.; Ruei-Yuan Wang; Hong-Sen WanPage(s): 609-612Digital Object Identifier 10.1109/IGARSS.2007.4422868

Satellite Observation Tracks Avian Flu12

An international, interdisciplinary team of researchers led by professor Xiangming Xiao of the University of New Hampshire is taking a novel scientific approach in an attempt to understand the ecology of the avian influenza, develop better methods of predicting its spread, and provide an accurate early warning system.

Accurate observation by remote sensing is, of course, only the first step. Its full importance is apparent when this novel source of information is incorporated into the decision-making process.Regardless of the field in which remote sensing is applied, the analytical procedure involved in transforming observation data into tools useful to decision-makers is fairly similar. Thus applying remote sensing in the analysis and management of business cycles follows an established analytical model. Here, a general chart presented by NASA may serve as guidance.

12 University of New Hampshire (2006, November 20). Satellite Observation Tracks Avian Flu. ScienceDaily. Retrieved July 22, 2008, from http://www.sciencedaily.com /releases/2006/11/061120101546.htm

21

Considering the wealth of research being carried out, it is surprising that business cycle economics has yet to tap the potential of earth observation through satellite remote sensing. This even harder to understand when one considers that business cycles are at the heart of economics.

The task, then, is to develop analytical approaches that will allow remote sensing viz. satellite observation to be applied to the observation and analysis of business cycles. Accordingly, the concepts underlying economic business cycle analysis must first be evaluated for their suitability to include remote sensing data.

22

4 TWO APPROACHES TO BUSINESS CYCLE ANALYSIS

4.1. THE BC-INDICATOR APPROACH

4.1.1. OVERVIEW

Economic activity has never developed along linear paths. Cyclical behavior is the pattern typical of economic development. Although the cyclical nature of economic activity has been recognised since the 17th century, modern analysis of business cycles (BC) only began at the start of the 20th century, (the well-known Harvard Barometer), and was very quickly faced with its own dramatic failure since the indicators were unable to forecast the world economic crisis of the 1920s. The results of the search, from that point on, for statistical business cycle indicators (BCI)13 could fill entire libraries. The general phenomena of a business cycle can easily be described as “…expansions occurring at about the same time in many economic activities, followed by similarly general recessions, contractions and revivals which emerge into the expansion phase of the next cycle”14

As has been mentioned earlier, the indicator approach to analyzing cycles of economic activity mainly involves interpreting statistical indicators of the underlying cyclical phenomena. The indicators have previously undergone an intensive screening process. The process of selecting hundreds of these indicators, over decades, has led to a deeper understanding of their behaviour over the duration of the cycles. That led to a bundling of indicators into groups with similar

13 One outstanding example of BCI-research is Geoffrey H. Moore, Ed. Business Cycle Indicators, NBER, Princeton 1961 within the realm of the National Bureau of Economic Research (NBER)

14 Arthur F. Burns and Wesley C. Mitchell, Measuring Business Cycles, 1946, p.21

23

behavior to produce time-pattern categories of indicators. Most often indicators are also aggregated to composite indices.

In applying indicators, probably the most difficult task is to identify the turning points in the cycles. This entails determining the duration of the respective cycle represented by the relevant indicator, either from peak to peak or trough to trough. It is through this process that the essential question of whether a recession is underway, or whether the cycle is in a phase of upswing, may be answered.

4.1.2. Examples

Because of their importance for economic analysis and decision-making, various catalogues of business cycle indicators have been established by a large number of statistical offices, research institutes of empirical economics and the like. A few of them from the U.S are mentioned here:

A relatively transparent set of indicators organized around time-patterns is provided by the Conference Board15. The basics of the underlying methods are to be found in the Conference Board’s Handbook16. The simple set of indicators looks as follows:

15 http://www.conference-board.org/economics/bci/16 http://www.conference-board.org/pdf_free/economics/bci/BCI-Handbook.pdf

24

Another example – taken from the National Bureau of Economic Research - gives insight into the delicate task of determining the duration, peaks and troughs of the cycles:

BUSINESS CYCLE REFERENCE DATES

DURATION IN MONTHSPeakTroughContractionExpansionCycleQuarterly datesare in parenthesesPeak

to TroughPrevious trough

to this peakTrough from

Previous TroughPeak from

Previous Peak

June 1857(II)October 1860(III)

April 1865(I)

25

June 1869(II)October 1873(III)

March 1882(I)March 1887(II)July 1890(III)

January 1893(I)December 1895(IV)

June 1899(III)September 1902(IV)

May 1907(II)January 1910(I)January 1913(I)

August 1918(III)January 1920(I)May 1923(II)

October 1926(III)August 1929(III)

May 1937(II)February 1945(I)

November 1948(IV)July 1953(II)

August 1957(III)

April 1960(II)December 1969(IV)November 1973(IV)

January 1980(I)July 1981(III)

July 1990(III)March 2001(I)December 1854 (IV)

December 1858 (IV)June 1861 (III)

December 1867 (I)December 1870 (IV)

March 1879 (I)

May 1885 (II)April 1888 (I)May 1891 (II)June 1894 (II)June 1897 (II)

December 1900 (IV)August 1904 (III)

June 1908 (II)

26

January 1912 (IV)December 1914 (IV)

March 1919 (I)July 1921 (III)July 1924 (III)

November 1927 (IV)March 1933 (I)

June 1938 (II)October 1945 (IV)October 1949 (IV)

May 1954 (II)April 1958 (II)

February 1961 (I)November 1970 (IV)

March 1975 (I)July 1980 (III)

November 1982 (IV)

March 1991(I)November 2001 (IV)--

128Average, all cycles:1854-2001 (32 cycles)1854-1919 (16 cycles)1919-1945 (6 cycles)

1945-2001 (10 cycles)  67* 31 cycles

** 15 cyclesSource: NBER

The most recent decision of the Business Cycle Dating Committee of the National Bureau of Economic Research is the determination that the last contraction ended in November 2001.

..

Announcement dates:

The November 2001 trough was announced July 17, 2003.The March 2001 peak was announced November 26, 2001.The March 1991 trough was announced December 22, 1992.The July 1990 peak was announced April 25, 1991.The November 1982 trough was announced July 8, 1983.The July 1981 peak was announced January 6, 1982.The July 1980 trough was announced July 8, 1981.The January 1980 peak was announced June 3, 1980.

Other Related Press Releases:

December 21, 1990, December 31, 1979, October 25, 1979, July 27, 1979

27

The NBER does not define a recession in terms of two consecutive quarters of decline in real GDP. Rather, a recession is a significant decline in economic activity spread across the economy, lasting more than a few months, normally visible in real GDP, real income, employment, industrial production, and wholesale-retail sales. For more information, see the latest announcement on how the NBER's Business Cycle Dating Committee chooses turning points in the economy and its latest memo, dated 07/17/03.

Source:

Public Information OfficeNational Bureau of Economic Research, Inc. 1050 Massachusetts Avenue Cambridge MA 02138 USA

617-868-3900

Recent data on BCI can be found in the WSJ - Wall Street Journal

e.g.: October 16, 2008 07:58 p.m. ET

Date ET Release For Actual Briefing.com Consensus Prior Revised From

Oct 16 08:30 a.m. Core CPI Sep 0.1% 0.1% 0.2% 0.2%

Oct 16 08:30 a.m. CPI Sep 0.0% 0.0% 0.1% -0.1%

Oct 16 08:30 a.m. Initial Claims 10/11 461K 475K 470K 477K 478K

Oct 16 09:00 a.m. Net Foreign Purchases Aug $14.0B NA $30.0B $8.6B $6.1B

Oct 16 09:15 a.m. Capacity Utilization Sep 76.4% 78.0% 78.0% 78.7%

Oct 16 09:15 a.m. Industrial Production Sep -2.8% -0.8% -0.8% -1.1%

Oct 16 10:00 a.m. Philadelphia Fed Oct -37.5 -5.0 -5.0 3.8

Oct 16 11:00 a.m. Crude Inventories 10/11 5611K NA NA 8123K

Oct 17 08:30 a.m. Building Permits Sep 845K 840K 854K

Oct 17 08:30 a.m. Housing Starts Sep 880K 870K 895K

Oct 17 10:00 a.m. Mich Sentiment-Prel. Oct 68.0 65.0 70.3

Oct 20 10:00 a.m. Leading Indicators Sep NA -0.3% -0.5%

Oct 22 10:35 a.m. Crude Inventories 10/18 NA NA NA

Oct 23 08:30 Initial Claims 10/18 NA NA NA

28

a.m.

Oct 24 10:00 a.m. Existing Home Sales Sep NA 4.93M 4.91M

Oct 27 10:00 a.m. New Home Sales Sep NA NA NA

Oct 28 10:00 a.m. Consumer Confidence Oct NA NA NA

Oct 29 08:30 a.m. Durable Orders Sep NA NA NA

Oct 29 10:35 a.m. Crude Inventories 10/25 NA NA NA

Oct 29 02:15 p.m.

FOMC Policy Statement

Oct 30 08:30 a.m. Chain Deflator-Adv. Q3 NA NA NA

Oct 30 08:30 a.m. GDP-Adv. Q3 NA NA NA

Oct 30 08:30 a.m. Initial Claims 10/25 NA NA NA

Oct 31 08:30 a.m. Employment Cost Index Q3 NA NA 0.7%

Oct 31 08:30 a.m. Personal Income Sep NA NA NA

Oct 31 08:30 a.m. Personal Spending Sep NA NA NA

Oct 31 09:45 a.m. Chicago PMI Oct NA NA NA

Oct 31 10:00 a.m. Mich Sentiment-Rev. Oct NA NA NA

Nov 03 10:00 a.m. Construction Spending Sep NA NA NA

Nov 03 10:00 a.m. ISM Index Oct NA NA NA

Nov 04 12:00 a.m. Auto Sales Oct NA NA NA

Nov 04 12:00 a.m. Truck Sales Oct NA NA NA

Nov 04 10:00 a.m. Factory Orders Sep NA NA NA

Nov 05 08:15 a.m. ADP Employment Oct

Nov 05 10:00 a.m. ISM Services Oct NA NA NA

Nov 05 10:35 a.m. Crude Inventories 11/01 NA NA NA

Nov 06 08:30 a.m. Initial Claims 11/01 NA NA NA

Nov 06 08:30 a.m. Productivity-Prel Q3 NA NA NA

Nov 07 08:30 a.m. Average Workweek Oct NA NA NA

Nov 07 08:30 a.m. Hourly Earnings Oct NA NA NA

29

Nov 07 08:30 a.m. Nonfarm Payrolls Oct NA NA NA

Nov 07 08:30 a.m. Unemployment Rate Oct NA NA NA

Nov 07 10:00 a.m. Pending Home Sales Sep

Nov 07 10:00 a.m. Wholesale Inventories Sep NA NA NA

Nov 07 03:00 p.m. Consumer Credit Sep NA NA NA

Nov 12 10:35 a.m. Crude Inventories 11/08 NA NA NA

Nov 13 08:30 a.m. Initial Claims 11/08 NA NA NA

Nov 13 08:30 a.m. Trade Balance Sep NA NA NA

Nov 13 02:00 p.m. Treasury Budget Oct NA NA -

$55.6B

Nov 14 08:30 a.m. Export Prices ex-ag. Oct NA NA NA

Nov 14 08:30 a.m. Import Prices ex-oil Oct NA NA NA

Nov 14 08:30 a.m. Retail Sales Oct NA NA NA

Nov 14 08:30 a.m. Retail Sales ex-auto Oct NA NA NA

Nov 14 10:00 a.m. Business Inventories Sep NA NA NA

Nov 14 10:00 a.m. Mich Sentiment-Prel. Nov NA NA NA

Nov 17 08:30 a.m.

NY Empire State Index Nov NA NA NA

Nov 17 09:15 a.m. Capacity Utilization Oct NA NA NA

Nov 17 09:15 a.m. Industrial Production Oct NA NA NA

Nov 18 08:30 a.m. Core PPI Oct NA NA NA

Nov 18 08:30 a.m. PPI Oct NA NA NA

Nov 18 09:00 a.m. Net Foreign Purchases Sep NA NA NA

Nov 19 08:30 a.m. Building Permits Oct NA NA NA

Nov 19 08:30 a.m. Core CPI Oct NA NA NA

Nov 19 08:30 a.m. CPI Oct NA NA NA

Nov 19 08:30 a.m. Housing Starts Oct NA NA NA

Nov 19 10:35 a.m. Crude Inventories 11/15 NA NA NA

30

Nov 19 02:00 p.m. FOMC Minutes Oct

29

Nov 20 08:30 a.m. Initial Claims 11/15 NA NA NA

Nov 20 10:00 a.m. Leading Indicators Oct NA NA NA

Nov 20 10:00 a.m. Philadelphia Fed Nov NA NA NA

Calendar Key:Actual refers to the actual figures after their release.Briefing.com refers to Briefing.com's forecast.Consensus represents the market consensus estimate for each indicator.Prior represents the last actual for each indicator. In cases where the release is a revision to an earlier estimate, as is possible with GDP, productivity, and U of Michigan sentiment, the last actual refers to the preliminary estimate for the same period. After a report is released, the Prior column reflects the prior figure as revised.The Revised From column lists the prior number as it was originally reported, ie before revision.Not included: Mitsubishi and Redbook chain store indexes are released every Tuesday morning. M2 is released every Thursday at 16:30 ET.

Despite the immense effort to raise the quality of BCI to the highest standards, the problem of timeliness persists. “…the NBER business cycle peak and trough dates are often determined with a substantial lag. For example, the March 1991 and November 2001 business cycle troughs were not announced by the NBER until nearly two years after the fact.”17 Speeding up the processing of statistics or applying methods of mathematical statistics are only second-best solutions in comparison to utilising the results of real-time observation.

4.1.3. The BCI approach and SOBI

Economic activity is such a complex phenomenon that even several hundred indicators may not suffice to describe it. Moreover, few of these indicators lend themselves to visual inspection. Needless to say, the fundamental determinants depend on financial factors. However, quite a few manifestations are observable: lower factory capacity is accompanied by lower thermal warming, decreases in trade imply less transportation, cuts in disposable personal income lead to lower demand for housing, bad weather destroys harvests etc. Thus, a few areas are accessible to visual inspection. Remote sensing should be able to provide the necessary tools to generate data which could supplement non-observable indicators with real-time information.

17 M. Chauvet and J. Piger, A Comparison of the Real-Time Performance of Business Cycle Dating Methods, Working Paper, University of Oregon, Feb 9, 2007, cf. also: M.W. Watson, Business Cycle Duration and Postwar Stabilization of the U.S. Economy, in: American Economic Review, 84, 1994, p. 24-46, or: Journal of Business Cycle Measurement and Analysis: Volume 3, No. 1, OECD, 2007

31

The methods to be used for combining BCI with SOBI will be described in general terms in chapter 5.1.

Remote sensing produces a multitude of statistical data. Even after selecting areas relevant to business cycle analysis not all of these data can be utilised. The process which has gone on in empirical business cycle economics throughout the past decades of establishing, describing, selecting and ordering business cycle indicators, must also occur for SOBI18.

Amongst the criteria which SOBI have to fulfill, the following are of particular importance:

Objectivity

As mentioned above, human interference may introduce bias into the production of BCI. SOBI should not suffer from this flaw.

Timeliness and continuous consistency

Timeliness is the overarching criterion for the use of BCI. SOBI are only useful if they reflect isochronically the phenomena observed. Unfortunately, disturbances in the atmosphere occur and can hamper observability. Frequent interruptions in observability would prevent the establishment of time series and interfere with the pattern of SOBI, making these data invalid for the analysis of BC.

Comparability with traditional statistics

Research has led to the development of several hundred business cycle indicators. It should therefore be no problem to identify an indicator used in official statistical methods comparable to the respective indicator generated by remote sensing.

Regionalization

SOBI are visual observations of events on earth. The geographical range of any set of data would be defined for any given observation. This, in turn, should be coordinated with the analytical business cycle model.

Classification

A last and conventional step in monitoring by SOBI would require these space-observed indicators to be ordered according to the established framework of leading, coinciding and lagging indicators.

18 Such tests are common whenever new insights and statistics for business cycles are under way, e.g.: “The Swiss Institute for Business Cycle Research regularly conducts business tendency surveys (BTS) amongst manufacturing firms. The information thus generated is available with a publication lead to the official Swiss sales, production, order and inventory statistics. It is shown that the survey data can be used to generate reasonably precise estimates of the reference series with leads of at least one quarter. Specifically, cross-correlations of quarterly series are computed to screen the data for pairs of highly correlated business tendency survey series and corresponding official statistics.” In: Journal of Business Cycle Measurement and Analysis 2003, vol. 2004, no. 1, pp. 116 - 141

32

The ultimate goal of generating and classifying SOBI would be to create a catalogue of such indicators comparable to the business cycle indicators described above, and where probably the best known guide is the work of NBER. It is also entirely possible that this new method of using visually observable phenomena could lead to new insights into business cycles.

One important qualification has to be underlined. Remote sensing is very costly. The volume of data users could acquire via SOBI is likely to be constrained by cost considerations. Commercially viable solutions need to be sought. It may also be expected that interest in tapping the potential of SOBI for BCI would be an incentive for SOBI providers to offer enhanced and competitively priced services.

33

4.2. ECONOMETRIC MODELS

4.2.1. THE BASICS

Econometric business cycle models are the most coherent analytical tools for investigating, forecasting and simulating decision-making in business cycles. Econometric models should also provide appropriately stringent guidelines for the integration of non-econometric methods such as SOBI into business cycle research. Thus considerable attention is paid to this crucial topic in this presentation. The following overview is not restricted to those features directly relevant to the integration of SOBI, but seeks to demonstrate the current state-of-the-art of econometric business cycle analysis, including models largely dependent on financial factors and thus less suitable for the integration of visually observable indicators. It is hoped that an overarching survey may suggest areas where data derived by SOBI analysis could be attached or integrated. To thoroughly explore potential practical applications of SOBI, it is necessary to undertake an extensive survey of models currently in use, even at the risk of overextending this chapter.

The peak period of large-scale econometric models, the 1960s, saw the development of an outstanding example: the Brookings-SSRC model. This ambitious model contains over 300 equations19 , and with the inclusion of separate modules could contain some 600 equations. A diagrammatic representation gives insight into the most important interactions:

19 James S. Duesenberry, Gary Fromm, Lawrence W. Klein and Edwin Kuh, Eds., The Brookings Quarterly Econometric Model of the United States, Rand McNally, 1965

34

FLOW DIAGRAM OF BROOKINGS-SSRC ECONOMETRIC MODEL

Applying this type of model revealed that in practice it was too large to provide useful analysis or economic forecasts. The need to down-size was evident. It has since become apparent that business cycles are increasingly determined by financial variables and public expenditures, and so the focus of more recent models is on these equations.

This is reflected in the following survey20 which describes the most important business cycle models.

The basic characterization of modern macroeconomic models is the paradigm by which a broadly neoclassical view of macroeconomic equilibrium coexists with a new Keynesian view of short-to-medium term adjustment. In addition, expectations of future values of endogenous variables, such

20 It is gratefully acknowledged that the major part of this survey was produced by the Austrian Institute of Economic Research, Vienna.

35

as exchange rates and inflation, are often important determinants of current behaviour, and their influence has been incorporated into macroeconometric models in various ways.

Subsequently, various macroeconometric models used by international organizations will be presented. These models can be considered as state-of-the-art models as they combine short-term Keynesian features with long-term neo-classical properties, drawing upon the rational expectations approach.

4.2.2. Disaggregated Dynamic Stochastic General Equilibrium (DSGE) models

General Equilibrium (GE) models are economies where agents buy, sell, and consume goods and securities in such a way that all markets clear. Market clearing is the situation when agents demand as many goods as are supplied (by other agents), which is associated with the market-clearing price. This equilibrium has many important aspects but probably the most important is the fact that the solution is efficient (welfare maximising) and can be achieved by a functioning price-mechanism. The general equilibrium approach has to be supplemented by Stochastic Dynamic Optimisation. Dynamic optimisation deals with the problem of calculating today’s optimal behaviour given certain beliefs about the future and future optimal behaviour. Stochastic dynamic optimisation works on uncertain future events. They not only affect the investment’s pay off but also future optimal behaviour. Forming rational expectations under such conditions remains a challenge and is even for very simple cases only possible by applying numerical approximations.This is being attempted by Computational General Equilibrium Models. CGE Models are often complex enough that they can be solved by numeric approximations only. Since analytically solvable problems are usually just simple text-book problems, generally all applied GE models are in fact CGE models.Finally, in order to come closer to reality, sectoral disaggregation is needed. This leads to Multi Sector Models.Real Business Cycle models were designed to explain the relation and co-movements of high-level or aggregate macroeconomic time series. They became famous and widely used because they served this purpose well, but they were not good at explaining co-movements across industrial sectors. Since these sectors are strongly interrelated through the production of intermediate goods, shocks to one sector should leave observable traces in other sectors. Understanding how responses follow a shock can also aid our understanding of how certain policies operate on the economic environment. Input-Output tables are a quantification of this network and can thus aid analysis of these mechanisms.

OECD – INTERLINK

36

The INTERLINK model of the OECD’s department for economics and statistics is used for analysing policies, and creating projections and simulations. Among its best known publications is the OECD Economic Outlook.

Each OECD member-country is represented by its own sub-model, and these are, in principle, separate, closed models.21 Non-OECD-countries are grouped into blocks, which are modelled in less detail. It is perhaps best to see INTERLINK as a framework, tightly connecting these many models into a larger whole through international trade.

The model has been continually extended by adding new sub-models and enhanced by new equations and variables. As already mentioned, these variables and equations are not necessarily the same for all members. Econometric studies are carried out, showing which variables and equations contribute to explanation power. Richardson (1988) presents a list in which the US, Japan, Germany, France, the UK, Italy and Canada each show up with around 250 equations and between 80 and 90 exogenous variables. Other member-states are modelled by around 130 to 170 equations with around 70 to 90 variables.

The equations can be split into three different groups. The first are the production functions, which calculate long-run output. With the exception of Japan, a constant-returns-to-scale Cobb-Douglas function is applied here. The second group of equations works on wages and prices. Since nominal rigidities are implemented, these equations mainly determine the speed of adjustment back to equilibrium after a shock. The third group, which may have up to 50 variables, consists of behavioural equations for consumption and import/export.

Working on the equations and variables is a necessary, but delicate, undertaking. Models and data showing the best fit during out-of-sample testing are sometimes theoretically unexplainable and vice versa. Thus there exists a trade-off between prognosis-quality and plausibility with respect to theory.

But as Richardson (1988) states, INTERLINK is primarily a prognosis tool. This pragmatic view also explains the high number of behavioural equations and empirical relationships, which have been under attack by the Lucas critique for decades now.

Concerning forecasting, the economists at the OECD confirm that the model is not supposed to substitute other mechanisms, but serves as a starting point for projections. The model’s results need to be improved on by international and country specific experts, who add data which is not used in the model itself, like a variety of economic indicators, surveys, stock-market assessments and many more.

The last big improvement of INTERLINK was a revision of the trade model. Apart from numerous important changes in the details (see Pain, Mourgane, Sédillot, and Le Fouler (2005)), there were two major changes implemented. Firstly, trade prices and volumes are now determined by behavioural equations with long-run equilibrium-corrections. The previous model had manufacturing as the main equation of foreign trade, which was calibrated instead of estimated.

21 Only Luxembourg and Belgium are combined in one model.

37

Secondly, non-linear deterministic trend functions are frequently used in both volume and price. Former versions of the model use non-linear functions only rarely, and with caution, for theoretical and practical reasons.

According to Dalsgaard, André, and Richardson (2001) the model does not feature forward looking expectations, which they admit is a drawback. This again exposes the model to the Lucas critique, since expectations have to be formed from past data, which is certainly not correct in the ever changing environment of the economic world. This is especially true if the effects of new policies are simulated. Real persons would instantly adapt their behaviour to the new circumstances, thus creating conditional expectations. The model does not do so, but simply extrapolates the past into the future. Conditioning on the new policy is therefore impossible, a task which is left to the experts.

The main economic set-up shows Keynesian features in the short-run and neo-classical properties in the long-run. Wages are subject to a bargaining process and expectations about inflation are adaptive. These aspects lead to frictions and lags, resulting in a sloping short-term Phillips curve. In the long-run, agents adapt and the relation between unemployment and inflation is nearly vertical. Some of the actual numbers given in Dalsgaard, André, and Richardson (2001) deviate a lot. The sacrifice ratio, stating how many percent unemployment has to be increased to decrease inflation by one percent, varies from 0.1 to 8.6. But these values have to be interpreted together with other elasticities, since they together determine the speed of adjustment of several variables on inflation or unemployment shocks.

Prices in the long term are modelled as marginal costs plus mark-up. Thus INTERLINK does not assume perfect competition, but allows prices to be set within a non-degenerate interval. Prices are also demand sensitive in the short term with delayed reaction to changes in costs.

Technological progress is disembodied and labour is supplied by a homogeneous labour-force without differences in skill or education.

European Commission – QUEST II

General Aspects

Traditional Keynesian models suffer from non-existing microfoundations. They describe the protagonists’ actions but they neither explain these actions nor do they allow the agents to adapt their behaviour to a new environment. Real Business Cycles (RBC) models are microbased by setting up the agents’ optimal decision problems. The New Keynesian Paradigm (NKP) merges the best of both ideas by introducing frictions and rigidities into RBC models to explain stickiness of prices and wages on the one side while avoiding the Lucas critique by using rational agents.

The European Commission’s QUEST II model is a model in the spirit of the NKP. It was one of the first to be implemented and used. According to Ratto, Röger, in‘t Veld, and Girardi (2005), it was

38

introduced so early that many parameters could not be estimated at that time, and it was necessary to calibrate them.

As can be expected, QUEST and QUEST II focus on the European Union and its most closely connected partners. Its main purpose is the analysis of economic policies and only secondarily forecasting, although good ex ante analysis requires a good out-of-sample fit.

The main difference between QUEST and QUEST II is a shift away from traditional Keynesian ideas towards a more supply-side point of view. Expectations became forward-looking, behaviour is modelled as optimal reactions to changes in the environment, and the supply side is now modelled in more detail. International trade is closed in such a way that trade balances and net foreign assets always sum to zero.

The long-term properties of QUEST II still closely resemble those of standard neoclassical models. Economic growth is driven by exogenous technological progress and population growth. One of the two main differences to pure neoclassical models is that companies act in the state of monopolistic competition and thus price their products with mark ups on marginal costs. These higher prices lead to lower demand and thus to lower economic activity than under perfect competition.

The second main deviation of long-term properties is involuntary unemployment. This is caused by the wage-bargaining process leading to rigidities which are never fully overcome.

Optimal consumption is determined by a Constant Relative Risk Aversion (CRRA) utility function, which has many convenient aspects, but also a few which are less positive. The most prominent drawback is probably the use of the risk aversion parameter as the determining variable for intertemporal substitution. Risk aversion and time preferences are therefore directly linked.

Households are interpreted as finitely lived, where in the end it is optimal to consume everything, and bequests to children have no utility. Optimal consumption is a function of inter alia probability of death, the real interest rate, the intertemporal elasticity of substitution, financial wealth (sum of total equity wealth, bonds and net foreign assets), and human wealth (present discounted value of the entire future stream of after-tax income). The assumption that people can and want to take out loans on their future income to finance present consumption is quite strong. Therefore the model also features so-called “liquidity constrained” agents, considering their current real disposable income as a source of consumption. Consumption smoothing is achieved by adapting the savings rate to expected income differences.

The production function is a nested CES and Cobb-Douglas function with capital, energy, and labour as inputs. Technological progress is labour augmenting and takes place at an exogenously given rate. Companies maximise the present value of their expected cash flow by changing output and mark ups. Thus total real output may differ from potential output, leading to a production gap, which plays an important part in the analysis of inflation. As already mentioned, mark ups lead to higher prices compared to perfect competition and thus reduce economic activity.

According to Tobin’s Q-model, investments are determined by the ratio of the company’s market value and the replacement value of its capital stock. Crucial factors in the size and frequency of

39

investments are expectations about future demand and costs as well as adjustment costs. The latter were estimated and found to be rather different among countries.

The labour market is defined in a quite detailed manner. Every quarter of the year, one quarter of the employees negotiate their wages at an individual level in a Nash-bargaining. Wages are fixed for one year. This introduces nominal rigidities, leading to the Keynesian aspects of the model. Important variables in the bargaining process are the relative bargaining strengths, labour income tax, unemployment benefits, the chances of finding or quitting jobs, and the vacancy costs for the firms, depending on the unemployment rate (labour market tightness).

Pricing occurs via mark up on marginal costs, which are partly affected by expected inflation. These expectations are mainly forward looking, but a fraction of the firms applies rules-of-thumb on past data.

Governments spend resources on unemployment benefits, goods, services, government wages, investments, transfers to households, and interest on debts. Revenues arise from labour income tax, corporate profit tax, value added tax, energy tax, and lump sum taxes. Rules are made to keep debt sustainable.

Capital is perfectly mobile across states, being subject to an endogenously determined exchange rate. Exchange rates’ values are calculated by no-arbitrage considerations, expected depreciation, and a risk premium. Monetary authorities are allowed to apply different policies and expectations about inflation.

As Ratto et al. (2005) describe, a new version of QUEST II is about to be developed and tested. Although the new Dynamic Stochastic General Equilibrium (DSGE) model is very similar to the two old versions in the long run, several important mechanisms were changed and nearly all parameters were re-estimated to fit actual data.

The biggest difference from a theoretical point of view is that households are no longer finitely lived. Thus a household is not interpreted as a person, but, as the name suggests, as households or families which continue to exist after the death of a single person. Although this might seem strange at the beginning, it describes much better the behaviour of many persons. Consumption was given a backward looking equation to allow for habit persistence. People do not want to give up a level of consumption already attained, even when doing so would be rational under standard dynamic optimisation.

The new model was not fully implemented in 2005 and had several key components missing, but the first results reported by Ratto et al. (2005) suggested further efforts should be supported.

Variables and Equations

Quest II uses a number of variables which are only partly observable or highly aggregated (such as the average price of consumer goods). Several variables are endogenous, like the savings rate or mark ups in pricing. It seems likely that not all variables or their use are fully described. The depreciation rate of capital is mentioned in the results, but does not show up in the model description, and corporate profit tax can be expected to be part of the pricing- or production block,

40

but it is only mentioned within government expenditures. The same holds true for transfers to households. These resources are spent by the government, but do not appear at their target groups, which would mean that these expenditures either disappear, or show no effect. Both alternatives are highly unlikely.

Table xx: Variables in QUEST II, their observability, and their use in the model's building blocks, strictly according to Röger and in't Veld (2002).

Variable dir. observ. Appearance *Human wealth yes cons.Bond investments yes cons.Net foreign assets yes cons.Equity wealth yes cons.Time preferences no cons.Probability of death yes cons.Intertemp. elast. subst. no cons.Real interest rate yes cons., fin.Price index yes cons., inv.Price of consumption goods no cons.Share of constrained consumers no cons.Real disposable income yes cons.Capital stock yes prod., inv.Energy yes prod.Employment yes prod., lab.Efficiency index of capital stock no prod.Labour augmenting technical progress no prod.Investment purchases yes inv.Installation costs no inv.Price of investment goods no inv.Opportunity costs of capital no inv.Relative bargaining strength of workers no lab.Labour income tax yes lab., govUnemployment benefits yes lab., gov.Labour productivity yes lab.Vacancy costs no lab.Labour market tightness yes lab.Frequency of price adjustment no pric.Share of backward looking firms no pric.Purchase of goods and services yes gov.Government wages yes gov.Government investments yes gov.Transfers to households yes gov.Interest rates on government debts yes gov.Corporate profit taxes yes gov.

41

Value added tax yes gov.Energy taxes yes gov.Debt-to-GDP target level yes gov.Risk premium on exchange rates no fin.Targeted inflation rate yes fin.

Source: Röger and in't Veld (2002).

*) cons.: Consumption, prod.: Production, inv.: Investments, lab.: Labour market, pric.: Pricing of products, gov.: Government, fin.: Financial markets.

Models of the Federal Reserve Board

General Aspects

The US Federal Reserve Board started by constructing two models, MPS (MIT, University of Pennsylvania, and Social Science Research Council) and MCM (Multi Country Model), in the late 1960s and mid 1970s.

As Brayton, Levin, Tyron, and Williams (1997) write, the design of MPS was done by Ando, Modigliani, and de Leeuw and was largely influenced by the then widely used large-scale macro models, starting with a set of 60 behavioural equations. The basic structure was based on the IS/LM-theory, augmented by a Phillips curve and a neo-classical growth model for the long term properties. Neutrality of money was emphasised, pricing was done by mark ups on marginal costs, and expectations were implemented too. According to the state of the art at that time, expectations were adaptive and backward looking. Since MPS was designed to be a policy analysis tool, many possibilities to interfere were implemented.

In 1996, the Federal Reserve Board officially introduced two new models, the domestic FRB/US and the international FRB/MCM or FRB/WORLD. The latter models 12 regions with 250 behavioural equations, out of which 40 represent the US. The large number of equations was chosen deliberately to allow a wide range of policies to be analysed. This leads directly to the problem of estimating all equations in parallel, which could not be done in 1997. Thus the model does not feature all the requirements for a general equilibrium model, but the necessary improvements were at least planned.

Expectations can be formed in a backward looking manner, or as model consistent/rational expectations, where adaptive expectations are calculated by a vector autoregressive process. Care is taken that tradable financial variables adjust instantaneously while prices, wages, and other such variables adapt slowly.

Variables

The publicly available specifications on the FRB-models are unfortunately not very precise, e.g. there is no indication whether a variable is en- or exogenous. Highly aggregated variables are indicated as unobservable since the resulting value depends very much on chosen weights. The

42

mutual assignments of variables and equations is also given just for some key elements in Brayton et al. (1997).

Table xx: Variables in FRB/US and FRB/WOLRD, their observability, and their use in the model's building blocks, strictly according to Brayton et al. (1997).

Variable dir. observ. Appearance *Corporate after tax cash flow noCost of capital yes dur., hous., inv.Dividends yesDepreciation rate yes inv.Stock of manufacturing and trade inventories yesCapital stock yesAbsorption price index (GDP+imports-government labour – inventories)

yes dur., wage

Price for business output noPrice non-farming business less housing + oil imports

no

Price crude energy yes price, wageRelative price of imports yesAfter tax profits yes dividendsShort-term interest rate yesCorporate bond rate yesProductivity trend no price, wageWeighted growth rate for employer social insurance taxes

yes

Terms of trade yesUnemployment rate yes price, wageTangible wealth yes cons., dur., hous.Business sector output yes inv.Non-farming business output yes workCompensation per hour yes priceMinimum wage yesDomestic sales yesDisposable income yes cons., dur., hous.Output gap probably cons., dur.Time trend no dur., hous., work

Source: Brayton et al. (1997).

*) cons.: Consumption, dur.: Motor vehicles and other durables, hous.: Housing, inv.: Inventory, work: Aggregated working hours.

IMF – MULTIMOD Mark III

43

MULTIMOD is a dynamic multi-country macro model of the world economy that has been designed to study the transmission of shocks across countries as well as the short-run and medium-run consequences of alternative monetary and fiscal policies.

The core model includes explicit country specific sub-models for each of the 7 largest industrial countries and an aggregate grouping of 14 smaller industrial countries. Two separate entities of transition economies and developing countries represent the remaining economies of the world.

A consistent theoretical structure is employed for all industrial economies, and cross-country differences are reflected by different estimates of parameter values. The model converges to a balanced growth path. The MULTIMOD modelling system includes a well-defined steady-state analogue model for each country and for the world economy as a whole. These steady-state models can be used to study the long-run effects of shocks that have permanent consequences for saving, capital formation, output, interest rates, exchange rates, and other variables. MULTIMOD also exhibits important short-term Keynesian dynamics that result from significant inertia in the inflation process. MULTIMOD assumes that behaviour is completely forward-looking in asset markets and partially forward-looking in goods markets.

The model has not been designed to be a forecasting tool. The baseline corresponds to the medium-term World Economic Outlook projections. These medium-term projections are then extended into a model-consistent balanced-growth path where the real interest rate is greater than the world real growth rate.

1. INFORGE

General Aspects

The German Gesellschaft für Wirtschaftliche Strukturforschung mbH (GWS) and Institut für Arbeitsmarkt- und Berufsforschung developed the INFORGE (Interindustry Forecasting Germany) model in the 1990s. Since then it has been subject to continuous extensions and enhancements. One of the most recent is the LÄNDER extension, which has independent models for the 16 federal states of Germany (Distelkamp, Hohmann, Lutz, Meyer, and Wolter (2003)). The basic model is explained in Lutz, Distelkamp, Meyer, and Wolter (2003).

INFORGE is an econometric model based on I/O-tables, featuring each of the 59 sectors, including government, private households, non-profit institutions serving households (NPISHs), and (non-)financial corporations. Each of the sectors is explicitly structured, and the sum of them all aggregates to macroeconomic variables. The authors mention that with the exception of tax rates, labour supply, and international variables, most of the variables are endogenous and highly interdependent. This in turn leads to estimation problems, which were solved by applying OLS techniques. These are simple and robust enough to estimate all the variables in parallel. Diagnosis statistics and economic knowledge about signs and orders of magnitude are used to eliminate insignificant and nonsensical variables from equations.

Hohmann et al. (2003) emphasise that INFORGE does not assume perfect competition and therefore is not a GE model. Agents are also only allowed to use bounded rationality, which has its

44

up- and downsides. On the one hand, real people are hardly ever perfectly rational, not even approximately. On the other hand, deciding which aspects of perfect rationality are curtailed, and to what extent, requires ad-hoc assumptions.

An important aspect of INFORGE is its completely endogenous government budget behaviour. This was implemented to guarantee balanced finance accounts, but limits the possibilities of policy analysis.

To get around the purely demand driven mechanism of other models featuring I/O-tables, INFORGE uses a different strategy. First, producers set prices by mark-ups on their marginal costs. Given these prices, the customers decide their optimal demand, which in turn results in the production of the demanded goods.

Variables

Equations in INFORGE are not given in detailed form, in most cases only as y=f(x1, x2, …), but variables are given to a much larger extent. However, it is not clear how complete or detailed is the list of equations (105 given in Hohmann et al. (2003)). It seems likely that several variables appear more often than stated. As mentioned above, most variables, with the exception of taxes, labour supply and international variables, are endogenous.

Table xxx: Variables in INFORGE, their observability, and their use in the model's building blocks, strictly according to Hohmann et al. (2003).

Variable dir. observ. Appearance *Private consumption demand yes priv.Disposable income priv.Interest rate priv., int.Shares of utilisation purposes no priv.Index of utilisation purposes no priv.Index of consumer prices yes priv.10 year US treasury bond eff. yield (exogenous) yes priv., equ., int.Time trend no priv., gov., intDem.Consumption demand for commodities no priv.CPX bridge matrix no priv.Value added taxes yes priv., gov., equ.Other taxes on products yes priv., equ.Demand for trade & transport services no priv., gov.Subsidies for trade & transportation no priv.Consumption expenditures at basic prices no priv.Prices of consumer products yes priv., lab.Market prices of consumer products yes priv.Consumption expenditures NPISHs no NPISHsGPD yes NPISHs, gov.Market prices of consumer products NPISHs yes NPISHsBasic prices of consumer products NPISHs yes NPISHsSocial security benefits yes gov.Demand health benefits of people aged 65+ no gov.

45

Demand health benefits of working people no gov.Demand health benefits of children no gov.Price of social security goods, aggregated no gov.Expenditures social security goods, aggr. no gov.Government consumption yes gov.Market price of government consumption no gov.Equipment expenditure of industry sectors no equ.Inflation rate yes equ., lab.CDAX stock index yes equ.Gross production of industry sectors no equ., prodIm., lab.Capital stock of industry sectors no equ., VAI., lab.Replacement rate of capital in industry sectors no equ.Purchases/sales of other equipment & assets no equ.IAX bridge matrix no equ.Equipment prices no equ.Trade & transport of equipment no equ.Equipment at basic prices no equ.Equipment unit prices no equ.Prices of competing goods no equ.Market prices of equipment no equ.Construction expenditures yes con.Export demand of a commodity group no exp.Market prices of export goods no exp.Export at current market prices yes exp.Trade & transportation of export goods no exp.Subsidies on export goods exp.Export at basic prices no exp.Export at constant prices no exp.Unit costs of sector yes exp., finDem., VAI.Inventory stocks (exogenous) no finDem.Final demand at basic prices yes finDem.Final demand at market prices yes finDem., prodIm.Net commodity taxes no finDem., VA.Input coefficients between sectors yes intDem., VA.Relative prices of intermediate inputs yes intDem.Price of gross production of sector yes intDem.Intermediate inputs of commodity groups yes intDem., prodIm., VA.Imports yes prodIm.Demand for imports no prodIm.Price of import goods no prodIm.Value added of intermediate products no VAP.Manufacturing prices yes VAP.STX matrix VAP.Government commodities subsidies VAP.Price of intermediate inputs yes VAP.Gross value added yes VAP.GDP in constant and current prices yes VAP., lab.Price deflator for GDP yes VAP., lab.

46

Gross value added of industry sector VAI.Government subsidies yes VAI.Current investments yes VAI.Gross operating surplus no VAI.Average wage rate per hour yes lab.Total number of employees yes lab.Unemployment rate yes lab.Annual working time per employee (exogenous) yes lab.Total annual wage yes lab.Sum of gross wages and salaries in sectors yes lab.Social security contribution rate yes lab.Payments for social insurances yes lab.Expenditures for social security benefits yes lab.Retirement payments & unemployment benefits yes lab.Revenue of the environmental tax reform yes lab.Sum of gross wages and salaries for households yes lab.Labour costs per employee yes lab.Labour demand of sectors no lab.Number of self employed persons in sectors yes lab.Productivity of labour per employee yes lab.Number of employees German nationals yes lab.Labour force potential (exogenous) yes lab.Job creation measures (exogenous) yes lab.Labour force yes lab.Labour force reserve yes lab.Base refinancing rate of the ECB (exogenous) yes int.10 year US treasury bond eff. rate (exogenous) yes int.Interest rate for consumer credits yes int.

Source: Hohmann et al. (2003). *) priv: Private consumption, NPISHs: Consumption of non-profit Institutions serving households, gov.: Consumption expenditures of the general government, equ.: Equipment investment, con.: Construction expenditures, exp.: Export demand, finDem.: Aggregated final demand, intDem.: Intermediate demand, prodIm.: Domestic production and import, VAP.: Gross value added of the production sector, VAI.: Gross value added of the industries and its components, lab.: Labour market, int.: Interest rates.

4.2.3. Econometric sub-models prone to SOBI affiliation

Modelling of those sectors of the economy in which SOBI could perhaps be most easily integrated, like transport, energy or emissions, is very rare within the established field of econometric business cycle models. Although there are GE models featuring an energy or transportation sector, (like the

47

multi-sector model described below in Steininger, Friedl, and Gebetsroither (2006)), overall forecasting models do not include detailed submodels for such aspects. The main reason for this is the fact that any submodel raises numerous additional estimation problems. Including just one additional variable in a model increases the model's complexity and thus estimation complications (see e.g. Brayton et al. (1997) who were not able to estimate the model as a whole but are reported to have plans for estimating model blocks in the future.) This is why almost all forecasting models have at most a rudimentary set of equations for these sectors, or even treat them as exogenous. In the latter case, special forecasting models are applied and the results used in the macro model.

One such special model was introduced recently by Brännlund, Ghalwash, and Nordström (2007). Their approach is one of the few which does not only research direct demand for energy (heating, transport), but includes all kinds of goods and services, including recreational services. Their specific goal was to research the effects of technical progress and economic growth on emissions of carbon dioxide, sulphur dioxide and nitrogen oxide. To do so, they developed a model based on households deciding their consumption of heating, public/private transport, food, electricity, oil, and several more categories. Each category has its own energy demand and gas emission values, which sum up to the aggregated macro-values. Households make their decision on a cost-minimisation basis, where demand and prices influence each other. Technological progress increases the energy efficiency of certain services, lowering energy demand and thus energy prices in the first round. The second round effect is threefold, since lower energy prices of the first round

increase demand for energy-intensive services. Thus the effect of technological progress is partly reduced;

increase real income, which in turn increases the demand for all other goods and services, again raising energy demand and gas emissions;

change relative prices between all goods, shifting demands in non-trivial ways.

All three second-round effects are combined in the so-called rebound effect. The main questions in Brännlund, Ghalwash, and Nordström (2007) are whether there is a positive or negative net effect of technological progress on emissions and how high CO2-emission taxes should be to keep emissions constant. The motivation behind the question on the net effect is the Environmental Kuznets Curve, which is an inverted U-shaped curve, stating that low-income countries have low emissions due to a dominant agricultural sector. Middle-income countries push the development of a wide variety of industries, leading to high emissions, and high-income countries are said to have low emissions again due to higher energy efficiency. The results of Brännlund, Ghalwash, and Nordström (2007) show that the net effect for Sweden is negative, mainly because the price effect on real income greatly increases demand for other goods. Emission taxes should be considerably higher (up to more than twice their current level) to keep emissions constant. Since transportation is one of the key contributors to energy demand, another focus of this model is public and private transportation demand.

The econometric part of the model is based on the Almost Ideal Demand (AID) model,

48

where w is the weight of goods and services in the consumption decision, x is the total expenditure on non-durables, p is the group price of one of the four major groups (food, transportation, heating, other), r=1,..,4 is the group decided upon, and s=1,…,4 are the indices of all four groups. Stone’s price index ln P= w* ln p is used as index. Subgroups of all four main groups are treated in the same way. The rest of this model elaborates on elasticities.

Another recent model in a similar spirit is that of Kratena, Meyer, and Wüger (2008). They do not answer a specific question as in Brännlund, Ghalwash, and Nordström (2007), but are interested primarily in the estimation results. One of the advantages from an economic point of view is that, here, agents have a utility function, used to find optimal consumption of services and goods.

The paper of Steininger, Friedl, and Gebetsroither (2006) is an interesting application of a CGE-model showing the high degree of flexibility such models offer. The authors constructed a 35-sector non-passenger-transport-goods model based on the Austrian I/O table to study the effects of road pricing. Consumption is CES-driven, elasticities of substitution between transport- and non-transport-goods are calibrated, and transportation is split into public and private traffic, where the latter shows fixed and variable costs. Four different households were identified to represent income

quartiles. The model is closed by foreign exchange, emphasising the open Austrian economy.

4.2.4 MACROECONOMETRIC MODELS OF BUSINESS CYCLES AND SOBI

The above overview clearly illustrates that econometric models are the most sophisticated tools available to assess business cycles. Even when using only BCI, it is these formal models which provide the necessary analytical rigour. However, in contrast to past practice, the overview also demonstrates that the econometric models presently in use are overwhelmingly based on financial and monetary variables. They therefore do not lend themselves easily to variables open to visual inspection, such as those to be gained by remote sensing. Thus it is not a simple matter to establish links between the equations of these highly aggregated econometric models and SOBI applications.

Basically, there are two ways in which SOBI can be applied to current business cycle analysis using econometric models:

- First, detect such equations where the explanatory power would be increased by adding SOBI-derived variables. However, such equations are rare since econometricians strive to explain business cycles with as few equations as possible. Areas where equations could be enhanced by SOBI applications might include: Consumer functions, especially where

49

consumer durables are to be explained; Investment functions where disaggregated investment would include investment in non-residential construction; Macroeconomic supply functions where the calculation of potential output is at stake; Hybrid functions of value-added disaggregated by industrial sectors where sectors like transport or agriculture appear in the input/output part. Equations on foreign trade might also be examined.

- Second, build disaggregated sub-modules. Such stand-alone sub-models would then have to be joined to the main macroeconometric model. Areas where such sub-models may be applicable are similar to those just mentioned for equations.

In addition, it is essential that the geographical dimension of SOBI be kept in mind. Variables gained by remote sensing typically have to be displayed on geographical surfaces. All data acquired has a regional dimension. Thus, an econometric approach which is both highly disaggregated and regionalized is required. SOBI are applicable only to econometric models of specific areas. Areas which are particularly promising will be presented in chapter 6.

50

5 LINKING REMOTE SENSING TO TERRESTRIAL OBSERVATION: PROCEDURAL STEPS

5.1. CONVERSION OF SPATIAL DATA INTO STATISTICAL DATA

Remote sensing provides information which cannot be integrated directly into economic analysis. Images must first be converted into data, which is then fed into business cycle analysis. The following steps are necessary:

Transform the satellite images into data (statistical time-series?)

Match these data from remote sensing with attribute data, i.e. BCI (including testing the coherence of SOBI based time-series with conventionally produced BCI.)

Either enrich a conventional business cycle analysis framework using BCI with SOBI, or select equations viz. modules of econometric business cycle models with SOBI.

Choose sectors specifically suitable for applying SOBI.

Exploiting remote sensing data, converting them into usable data for data-banks and integrating them into specific models is already virtually a standard process in utilising space derived observations of earth. The uses of ESA’s Envisat22 can be cited as a guide to solving similar tasks in business cycle economics:

“Earth Observation data from Envisat are being used by United Nations agencies, the European Commission, non-governmental organisations, national and local authorities, and commerce.

Continuous and coherent global data sets are needed by the scientific and application community in order to understand better climatic processes and to improve climate models.

Some global applications require near real time data delivery (from a few hours to one day from sensing). Specific examples include: 22 http://envisat.esa.int/

51

forecasting of the sea state conditions at various scales; monitoring of sea surface temperature; monitoring of some atmospheric species (e.g., ozone for warning purposes); monitoring of some atmospheric variables (e.g., temperature, pressure, and water vapour, cloud top height, earth

radiation budget, etc.); monitoring of ocean colour for supporting fisheries and pollution monitoring (complementing the regional

mission).

Some of the global objectives require products available in off-line mode (days to weeks from sensing). Specific examples include quantitative monitoring of:

radiate processes; ocean-atmosphere heat and momentum exchange; interaction between atmosphere and land or ice surfaces; composition of the atmosphere and associated chemical processes ocean dynamics and variability; ice sheet characteristics and sea ice distribution and dynamics; large-scale vegetation processes in correlation with surface energy and water distribution; primary productivity of oceans; natural and man-made pollution over the oceans. support to large international programmes (GCOS, IGBP, etc.).

The ESLs, with industrial support, are responsible for producing the detailed algorithm specifications for Level 1b and Level 2 products, as well as for providing continuous support throughout the mission exploitation phase.

Phase I Product definition, algorithm development, and prototyping phasePhase II Support to the operational implementation and pre-launch verification phasePhase III Product validation phase, including in-orbit commissioningPhase IV Product quality control and product enhancement during the mission exploitation phase

The results of the ESL work is contained in a set of algorithm specifications:

Input/Output Data Definition defining the detailed content and format of all products, including auxiliary products

Detailed Processing Model and Parameter Data List specifying in detail the mathematical algorithms to produce the Level 1b and Level 2 products

Computer Resource Requirements

The development of the PDS operational processor is based on these applicable requirement specifications.

Instrument ESL/Industry Responsibility Deliverable

RA-2/MWR CLS (F) Ocean and Ice retracking Product and Algorithm Spec/Prototype

MSSL (UK) Ice and sea-ice retracking Product and Algorithm Spec/Prototype

52

MIPAS IROE (I), Univ. of Bologna (I), IMK/KfK (D), Lab. Physique Moleculaire (F) Global Fit inversion algorithm definition and scientific prototyping, physical/mathematical optimisations Algorithm Definition and Scientific Prototype

Oxford University (UK) Reference Forward Model and validation of inversion algorithm MIPAS Forward Model

IAA Granada (E) Study of non-LTE effects Database

Dornier (D) Detailed documentation and prototyping Product and Algorithm Spec/Prototype

DLR (D) Review of NRT algorithm and off-line enhancements Documentation

MERIS Freie Universitat Berlin (D) Cloud top height and water vapour Algorithm Theoretical Basis Document

Universite du Littoral (F) Atmospheric corrections over land Algorithm Theoretical Basis Document

Laboratoire Physique et Chimie Marine (F) Ocean colour algorithm Algorithm Theoretical Basis Document

GKSS Forschungscentrum Geesthacht (D) Coastal water Algorithm Theoretical Basis Document

Plymouth Marine Lab (UK) Coastal Water Algorithm Theoretical Basis Document

53

Institute of Ocean Science (CDN) Coastal water consultancy support

ACRI (F) Specification and prototyping Product and Algorithm Spec/Prototype

GOMOS Finnish Meterological Institute (SF)Level 1b and 2 processing definition Algorithm Specification

Service d'Aeronomie (F) Level 1b and 2 processing definition Algorithm Specification

IASB (B)Level 1b and 2 processing definition Algorithm Specification

ACRI (F) Specification and prototyping Product and Algorithm Spec/Prototype

AATSR RAL (UK)

Product and Algorithm Spec

DoE (UK)

Prototype

SCIAMACHY DLR-DFD (D)

Product and Algorithm

This is not the place to discuss the vast variety of formal methods in the field of remote sensing by which visual information is managed, ranging from pattern detection, clustering, texture analysis, spectral unmixing, regression and discriminance analysis, buffering, topological analysis etc. Even

54

the term “geoinformatics”, “as a combination of remote sensing data, ground observations encoded to geological coordinates, and Geographic Information System (GIS) data”23 , is insufficient to describe fully the formal methods currently in use.However, the principles for securing the coherence of space-derived data with ground data follow a rather simple scheme. The procedure runs as follows (with housing as the chosen example): Select the best sensors for observing housing construction within a predetermined area/region Project the images produced of housing construction activities observed via satellite onto GIS maps convert visual information into data enhance with attribute data (e.g. income per head of neighbouring resident population, traffic, other locational factors) correlate with conventional statistics on housing construction integrate into the relevant equations of an econometric model of housing.

1) Select sensor

Satellite/Sensor Origin (A few examples)GEOEYE-1 USA

WORLDVIEW-1 USAWORLDVIEW-2 USAQUICKBIRD USAPLEIDES-HR Constellation of 2 spacecrafts France

IKONOS USAKOMPSAT-2 KoreaORBVIEW-3 USAFORMOSAT-2 Taiwan .....

2) Observation of building activities

23 www.gis.com

55

3) Convert visual information into data24

24 http://www.esri.com/software/arcgis/arcinfo/about/features.html

56

4) Enhance with attribute data

57

5) Correlate with conventional business cycle statistics

Continuing with the example of housing:

US Census Bureau: The Survey of Construction (SOC) produces monthly estimates of housing starts and completions. Census Bureau “field representatives” sample individual permits within a sample of permit offices. Then the builders or owners who took out the sampled permits are interviewed to obtain start and completion dates along with sale dates and characteristics such as size and number of bedrooms. In addition, within a sample of land areas where building permits are not required, field representatives drive all roads looking for new residential construction activity.

6) Integrate in relevant equations of econometric business cycle model

A basic demand equation for residential housing observed by satellite could look as follows:

ln(SOBIHSr,t ) = α+β ln(SOBIHSr, t-1)+ γ ln(Y/POPr,t-1)- δ ln(UCr,t-1)+z

58

where SOBIHS is the housing stock in region r at time t observed by satellite, Y/POP is the real disposable income per head, UC stands for the user costs of housing, i.e. mainly the mortgage rates and rate of inflation and z represents other explanatory (shift) factors.

These procedural steps however involve sometimes sophisticated techniques25.

5.2 COMMERCIAL SOLUTIONS ON THE MARKET

Technicalities aside, commercial applications of remote sensing meanwhile are that widespread that conversion of spatial data and attribute data can rely on commercially viable software26. ESRI – ArcGISis probably is one of the best developed. It comprises a geodatabase27 which provides a large array of functions.

In a slightly marketing-like fashion FME28 displays the conversion solutions of spatial data and their effective commercial use. For specific business cases this seems to be well commercially viable, thoughgh it remains to be seen whether it could fulfill the more demanding tasks of business cycle observation. In In any case the FME approach is interesting:

Nimbus programProject VanguardSEASATTOPEX/PoseidonTIROS

Retrieved from "http://en.wikipedia.org/wiki/Earth_observation_satellite"Categories: Earth observation satellites

Screen readers: Skip navigation

25 Cf.e.g. L. Van Genderen, C. Pohl J.Review article Multisensor image fusion in remote sensing: concepts, methods and applications, in: International Journal of Remote Sensing, Volume 19, Number 5, 1998 , pp. 823-854(32)

26 http://www.geospatial-solutions.com/geospatialsolutions27 http://www.esri.com/software/arcgis/geodatabase/index.html28 http://www.safe.com/products/desktop/overview.php

59

6 REMOTE SENSING AND BUSINESS CYCLE OBSERVATION

60

61

62

63

7 IMPLEMENTING REMOTE SENSING: BUSINESS CYCLE ANALYSIS IN APPROPRIATE SECTORS

7.1. The task

It has been argued above at length that the macroeconomic character of business cycles allows only limited use of visual monitoring with the aid of SOBI. Money supply, public budgets, personal disposable income, prices, taxes, to cite just a few important variables, cannot be directly visually observed. These factors can be usefully expressed only through statistics. On the other hand, there are a number of manifestations of economic activity where direct visual observation of the “steam” of the economic machine is clearly possible. The following areas may be considered particularly promising for remote sensing observation and constitute significant aspects of business cycles

7.2 HOUSING

7.2.1 Approaches to assessing the housing cycle

Business cycle analysis is in constant search of indicators which offer the following features: they should exhibit a specific time pattern, preferably in the form of indicating a lead in the cycle; they should not be irrelevant for the cycle, i.e. they must represent a certain weight in the components of the cycle; and, to serve the present purpose, they should be accessible to remote sensing without physical disturbances.

Housing seems to fulfill these criteria very well. Beyond the significance of housing in general as a BCI, it should be noted that investment in housing is one of the priority targets for business cycle

64

policy, working mainly through monetary instruments29. Housing therefore is of prime interest to the credit markets. The importance of housing and its impact on the economy has been demonstrated at the outset of this study, ie. in the recent “case study” of October 17th on housing.

It is important to note that in any survey of “housing” the term is often used very loosely. The distinction must be made from the outset whether reference is to residential or non-residential housing, the stock of (existing) houses or investment (as flow-figure) into housing, investment demand for housing or actual construction of houses, building permits or housing starts etc. For the present purpose of achieving timeliness, an explanation from the US Bureau of the Census is important30:

Relationship Between Building Permits,Housing Starts, and Housing Completions

We are frequently asked why the building permits, housing starts, and housing completions series do not match over time, both in total and by category. In response, we thought it would be helpful to list the main factors

causing differences and to estimate the contribution of each based on data from 1999 to 2004.Background -- How the Surveys WorkNew housing construction data are collected in two surveys:The Building Permits Survey (BPS) produces estimates of the number of permits issued for new housing units each month. This is done through a mail survey of a sample of permit offices. Permit offices not in the monthly sample report annual numbers at the end of each year. Monthly data for States, Regions, and the U.S. are weighted sample based estimates reflecting the total building permit universe.The Survey of Construction (SOC) produces monthly estimates of housing starts and completions. Census Bureau “field representatives” sample individual permits within a sample of permit offices. Then the builders or owners who took out the sampled permits are interviewed to obtain start and completion dates along with sale dates and characteristics such as size and number of bedrooms. In addition, within a sample of land areas where building permits are not required, field representatives drive all roads looking for new residential construction activity.Factors Affecting the Permit-Start-Completion Relationships1.         Starts and completions in non-permit areasHousing starts and completions estimates cover the entire United States, not just areas requiring building permits. The number of housing units built in non-permit areas is about 2.5 percent of the total. Nearly all are single-family houses.(Note that the number of jurisdictions (or “places”) requiring building permits increases over time as non-permit places become permit-issuing. The Census Bureau’s universe of permit offices was increased in 2005 from 19,000 to 20,000 places.)

29 Cf. E.g. Viggo Nordvik, Selective housing policy in local housing markets and the supply of housing , In: Journal of Housing Economics, Volume 15, Issue 4, December 2006, p. 279 ss.purchase

Selective programs and subsidies have an impact on both the financial position and the housing conditions of the household to whom they are allocated. They also affect the equilibrium outcome in housing markets. This study analyzes how the housing stock in Norwegian municipalities is affected by selective targeted interventions on the supply and demand-sides of the market. The empirical analysis shows that additions to the stock of public housing increases the total housing stock. For every 100 new public units built, 60 units are added to the total housing stock. Demand-side subsidies are also shown to increase the size of the housing stock. Using a linear spline it is shown that the magnitude of the marginal effect on the total size of the housing stock is strongly decreasing in program size.

30 http://www.census.gov/const/www/nrcdatarelationships.html

65

2.         Changes after the issuance of permitsMany times, changes to the status of buildings take place after the permit has been issued, affecting the permit to start/completion relationship.Reclassification -- Townhouses are classified as single-family houses according to Census definitions, however, permit offices frequently classify them as multifamily structures. In SOC, we often sample permits for multifamily buildings that our field representatives later determine are townhouses. This reclassification results in significantly more single-family starts and completions (and less multifamily) than are shown in the permit data.Abandons -- Construction is sometimes abandoned after permits are issued but before construction is started, affecting the permit-to-start relationship, or after construction is started, affecting the start-to-completion relationship. Abandon rates can fluctuate over time due to conditions in the economy.Design changes -- Builders also can make design changes after the issuance of the original permits. This is more common with the construction of apartment buildings where the final number of units may be more or less than originally planned.Misclassification -- Permit offices sometimes incorrectly classify permits as new residential construction when the permits are actually for home improvements, the setting up of mobile homes, or the construction of non-residential buildings. Census field representatives will subsequently “out-of-scope” these permits if sampled in SOC.

3.         Permit revisions not applied to starts and completionsPart of the calculation of housing starts and completions involves a procedure where estimates of monthly permit authorizations based on SOC sample cases are ratio-adjusted to the more complete estimate of permits based on the monthly BPS. However, monthly permit estimates from the BPS are subsequently revised at the end of each year when results of the annual survey are incorporated. Under current procedures, the final revised permit numbers are not used in the calculation of starts and completions. Over the past few years, final permit estimates have been about 1.5 percent higher than the preliminary permit estimates used to develop the starts and completions data. This difference should be smaller in 2005 as we have redesigned the monthly BPS sample, which was 10 years old in 2004.

4.         Change in inventories between time periodsIn comparing the numbers of permits, starts, and completions over time, changes in the level of two inventory figures must be taken into account. The number of units “authorized but not started” affects the relationship between permits and starts, and the number of housing units “still under construction” affects the relationship between starts and completions.

Summary of FindingsThe following estimates of the average effect of each factor on the relationship of permits, starts, and completions are based on a summary of data from 1999 to 2004. The estimates were derived by either comparing published estimates for recent years or by tabulating unpublished data. For the most part, they are rough approximations, and measurements of their sampling errors have not been calculated. Please note that the estimates shown here are based on sample surveys and subject to sampling variability as well as nonsampling error.”

The OECD’s statistics database is another source31. The relevant indicator is defined as follows and seems to be particularly suitable for monitoring by remote sensing:

“Work started > ConstructionData covers construction work start for dwellings or buildings (residential and non-residential building in construction industry).Data generally refer to number of dwellings for which construction work commenced in the reference period. In some cases data may refer to gross surface or physical volume of construction 31 http://stats.oecd.org/wbos/Index.aspx?querytype=view&queryname=93

66

work started for dwellings or buildings. Construction is considered as underway once a foundation has been laid or is in preparation, for example digging has begun.”

It should be noted that significant lags occur between actual events and the publication of national data by the OECD.

Analysis of housing during the business cycle is carried out on the basis of such statistics. As already described in this study, it is undertaken either in the form of the indicator approach by which a (set of) indicators reflect fluctuations in housing or with the aid of econometric models of housing.

The approaches taken in the analysis of “housing” within business cycles can be represented by the following chart:

Definition of “housing”, “construction, investment in housing…

Indicator approach of BC: Selecting adequate indicators Monitoring with regard to BC

Supplement with SOBI providing real time observation

Econometric model approach: Select adequate econometric housing model

Supplement with SOBI

Connect econometric housing sub-model to econometric BC model

Whether for the indicator or the econometric approach, developing, selecting and monitoring indicators in the field of construction/housing occupies an important place in economic research institutes.

6.2.1.1. Approaching the housing cycle via BC indicators

67

Business cycle indicators for housing can look back on a long tradition in BC research. A prominent example is the role of building permits serving as a leading indicator to construct the composite business cycle index of the Conference Board32 which has already been cited:

Another example33 can be taken from N. Roubini from NYU:

Housing Starts/Building Permits

Importance: ***

Definition: The housing industry accounts for about 27% of investment spending and 5% of the overall economy. Housing starts is important because it is a leading indicator. Sustained declines in housing starts slow the economy and can push it into a recession. Likewise, increases in housing activity triggers economic growth.

INCLUDEPICTURE "http://www.census.gov/indicator/www/img/c20_curr.gif" \* MERGEFORMATINET INCLUDEPICTURE

32 The Conference Board, Business Cycle Indicators, monthly reports33 www.stern.nyu.edu/~nroubini/bci/bci.html

68

"http://www.census.gov/indicator/www/img/c20_curr.gif" \* MERGEFORMATINET

  Related Indicators: Source: Bureau of the Census of the U.S. Department of Commerce Frequency: Monthly Availability: Two to three weeks following the reported month Volatility: Moderate Likely Impact on Financial Markets:                    Interest Rates: Larger-than expected monthly increase or increasing trend is considered inflationary, causing bond prices to drop and yields and interest rates to rise.                     Stock Prices: �.                     Exchange Rates: ....

Analysis of the Indicator: Housing data tracks the four major regions of the U.S.: Northeast, Midwest, South, and West. Building permit data is released at the same time as housing starts. Permit activity provides insight into housing and overall economic activity in upcoming months. It is so important that it is included in the index of leading economic indicators.

Housing activity is directly impacted by mortgage rates. Higher interest rates increase housing costs and reduce the number of qualified borrowers, thus a decline in home sales and drop-off in starts. Conversely, lower interest rates increase housing affordability and spur home sales and housing starts. Housing data can have a significant impact on the bond market. A stronger-than-expected report is viewed negatively, suggesting strong growth and possible inflationary side-effects. A weak report has the opposite effect on the market.

WEB Links A Graph of the latest Housing Starts data from The Economic Statistics Briefing Room of the White House. The latest Housing Starts report from BLS.  

Should there be need for any further proof of the importance of housing, consider the following quote34:

“Economic Indicators: Housing Starts

34 http://www.investopedia.com/university/releases/housingstarts.asp

69

By Ryan Barnes  Release Date:On or around the 17th of the monthRelease Time:8:30am Eastern Standard TimeCoverage:Previous month's dataReleased By:U.S. Census BureauLatest Release:http://www.census.gov/const/www/newresconstindex.html

BackgroundThe New Residential Construction Report, known as "housing starts" on Wall Street, is a monthly report issued by the U.S. Census Bureau jointly with the U.S. Department of Housing and Urban Development (HUD). The data is derived from surveys of homebuilders nationwide, and three metrics are provided: housing starts, building permits and housing completions. A housing start is defined as beginning the foundation of the home itself. Building permits are counted as of when they are granted.

Both building permits and housing starts will be shown as a percentage change from the prior month and year-over-year period. In addition, both data sets are divided geographically into four regions: Northeast, Midwest, South and West. This helps to reflect the vast differences in real estate markets in different areas of the country. On the national aggregates, the data will be segmented between single-family and multiple-unit housing, and all information is presented with and without seasonal adjustment.

Housing starts and building permits are both considered leading indicators, and building permit figures are used to compute the Conference Board's U.S. Leading Index. Construction growth usually picks up at the beginning of the business cycle (the Leading Indicator Index is used to identify business cycle patterns in the economy, and is used by the Federal Open Market Committee (FOMC) during policy meetings).

What it Means for InvestorsThis is not typically a report that shocks the markets, but some analysts will use the housing starts report to help create estimates for other consumer-based indicators; people buying new homes tend to spend money on other consumer goods such as furniture, lawn and garden supplies, and home appliances.The housing market may show the first signs of stalling after a recent rate hike by the Federal Reserve. This is because rising mortgage rates may be enough to convince homebuilders to slow down on new home starts. For investors looking to evaluate the real estate market, housing starts should be looked at in conjunction with existing home sales, the rental component of the Consumer Price Index and the Housing Price Index (also available from the Census Bureau). (For related reading, see Investing In Real Estate.)

According to the Census Bureau, "it may take four months to establish an underlying trend for building permit authorizations, five months for total starts and six months for total completions", so investors should look more closely at the forming patterns to see through often-volatile month to month results. 

Strengths

Very forward-looking, especially building permits; a good gauge for future real estate supply levels Can be used to identify business cycle pivot points Sample size covers approximately 95% of all residential construction in the U.S.

Weaknesses

No differentiation between size and quality of homes being initiated, only the nominal amount Only focuses on one area of the economy

The Closing LineHousing starts is best used as a business cycle indicator and a tool for investors researching the real estate markets.”

Actual recent data:

70

August Housing Starts

Commentary and background from Briefing.com. See an index of reports.

Updated: October 16, 2008 07:56 p.m. ET

Big Picture

The housing sector has been in a deep recession.  Fortunately, there are now some signs that the rate of decline is slowing, and even that some stabilization is occurring.  The rate of decline in existing home sales has slowed over the past half year.  Sales are not picking up, but a bottoming is preceded by a leveling off.  Now, housing starts and permits are starting to level off as well.  It may well be that the housing sector stabilizes over the summer months, and picks up in the third quarter.  Lower mortgage rates and a stabilizing economy will help.  Lower prices on homes will ultimately stimulate demand, but for now may inhibit sales as the urgency to buy is mitigated. The housing sector is a long way from anything that can be called a recovery, but even a general stabilization would help boost GDP numbers by eliminating what has been a major negative on the numbers the past year.

Category Aug Jul Jun May AprStarts 895K 954 1089 982 10041 Unit 630K 642 663 682 681Multi Units 251K 301 404 280 308Permits 854K 937 1138 978 982

In Europe, an overview of housing indicators can easily be gained from Euroconstruct35 and the

35 http://www.euroconstruct.org/ EUROCONSTRUCT provides detailed up-to-date construction market forecasts for all main construction sectors: residential construction, non-residential construction and civil engineering.These forecasts are presented at the biannual EUROCONSTRUCT-Conferences to a broad audience for the first time. Regular participants are mainly decision makers from the construction sector and real estate business, as well as from related markets (such as suppliers or from the financial and public sector).

71

participating economic research institutes36. The main indicators37 provided in the reports include: “ CONSTRUCTION BY TYPE [2007: volume in mill. euro, % change in the period analysed] Residential Construction - New / Renovation / Total Non-Residential Construction - New / Renovation / Total Building - New / Renovation / Total Civil Engineering - New / Renovation / Total Total Construction Output - New / Renovation / Total Domestic Cement Consumption - New / Renovation / Total”

Having examined cyclical housing indicators in general, the ways in which SOBI can improve monitoring of housing cycles needs to be clarified.

The appropriate route is via GIS, Geographic Information Systems. In fact, GIS has already acquired a noteworthy reputation in the fields of real estate and housing research38. It has begun to be widely used in commercial applications. Real estate transactions rely on timely visual information about property location and the like. The necessary software is available on the market. However it mostly consists only of “images” of the real estate site matched with additional information like topographic items (roads...), appraisals, demographics, taxes, finance etc.

The process of matching GIS Data with SOBI is carried out with the help of techniques described in chapter 5.1.

The subsequent step, then, involves assessing the observability of construction (as a precondition of residential investment decisions). The abundance of data available for visual monitoring of construction is overwhelming39:

Overview: Satellite sensors providing visual inspection of construction activities

36 A sample report for Austria can be downloaded: http://www.euroconstruct.org/publications/cr_sample.pdf37 Cooperation of member institutes is assured by AQUIEC, the Association for the Quality of the Economic Indicators of

the Construction Industry38 G. I. Thrall, GIS Applications in Real Estate and Related Industries, in: Journal of Housing Research, Vol. 9 Iss. 1, 1998 p.

65 ss., L. Anselin, GIS Research Infrastructure for Spatial Analysis of Real Estate Markets, in: Journal loc. cit. p. 77 ss.39 Fernerkundung in der Konjunkturforschung, GeoVille Information Systems GmbH, Bericht an FFG, Innsbruck, 12 01 2008

72

Satellite/

Sensor

Origin

Orbit

Inclination

Height

Spectral Bands

Spatial Resolution

Panchromatic (Pan)

Multispectral (MS)

Revisit Time

Image Swath

Maximum Deviation from Nadir

Dynamic Range

Stereo Capability

Launch and expected (or completed) Lifetime

Current Status (11/2007)

GEOEYE-1

USA

Polar sun-synchronous

98 degrees

684 km

Pan: 450-900 nm

Blue: 450-520 nm

Green: 520-600 nm

Red: 625-695 nm

NIR: 760-900 nm

Nadir:

Pan: 0.41m

MS: 1.65m

1-3 days, depending on latitude and viewing angle

Nadir:

15.2 km

60°

11 bits per pixel

Across track

and

along track

2008

Operational life time 7 years

Expected life time 10 years

WORLDVIEW-1

USA

Polar sun-synchronous

97.2 degrees

496 km

Pan: 400-900 nm

Nadir:

0.5 m

20° off nadir:

0.55 m

1-2 days, depending on latitude and viewing angle

Nadir:

17.6 km

40°

11 bits per pixel

Across track

and

along track

2007

Operational life time 7 years

WORLDVIEW-2

USA

Polar sun-synchronous

97.2 degrees

770 km

Pan: 450-800 nm

Blue: 450-510 nm

Green: 510-580 nm

Red: 630-690 nm

NIR1: 770-895 nm

Plus:

Coastal: 400-450 nm

Yellow: 585-625 nm

Red edge: 705-745 nm

NIR2: 860-1040 nm

Nadir:

Pan: 0.46 m

MS: 1.8 m

20° off nadir:

Pan: 0.52 m

MS: 2.4

1-4 days,

depending on latitude and viewing angle

Nadir:

16,4 km

40°

11 bits per pixel

Across track

and

along track

2008

Operational life time 7.25 years

QUICKBIRD

USA

Polar sun-synchronous

97.2 degrees

450 km

Pan: 450-900 nm

Blue: 450-520 nm

Green: 520-600 nm

Red: 630-690 nm

NIR: 760-900 nm

Nadir:

Pan: 0.61m

MS: 2.44m

25° Off-Nadir: Pan: 0.72m

MS: 2.88

1-3.5 days, depending on latitude (30° off-nadir)

Nadir:

16.5 km

45°

11 bits per pixel

Across track

and

along track

2001

7 years

Active

73

PLEIDES-HR

Constellation of 2 spacecrafts

France

Polar sun-synchronous

98.2 degrees

695 km

Pan: 480-830 nm

Blue: 430-550 nm

Green: 490-610 nm

Red: 600-720 nm

NIR: 750-950 nm

Nadir:

Pan: 0.7 m

MS: 2.8 m

5 day with one satellite,

4 days with 2 satellites

Nadir:

20 km

30°

12 bits per pixel

Across track

(simultaneous stereoscopic acquisition mode)

2009

5 years

IKONOS

USA

Polar sun-synchronous

98.1 degrees

680 km

Pan: 450-900 nm

Blue: 450-520 nm

Green: 510-595 nm

Red: 630-700 nm

NIR: 760-850 nm

Nadir:

Pan: 0.82m

MS: 3.2m

26° Off-Nadir: Pan: 1m

MS: 4m

Ca. 3 days at 40° latitude

Nadir:

11.3 km

26° Off-Nadir:

13,8 km

45°

11 bits per pixel

Across track

and

along track

1999

7 years

Active

KOMPSAT-2

Korea

Polar sun-synchronous

98.1 degrees

685 km

Pan: 450-900 nm

Blue: 450-520 nm

Green: 520-600 nm

Red: 630-690 nm

NIR: 760-900 nm

Nadir:

Pan: 1 m

MS: 4 m

3 days Nadir:

15 km

56°

10 bits per pixel

Across track

and

along track

2006

3 years

Active

ORBVIEW-3

USA

Polar sun-synchronous

97.3 degrees

470 km

Pan: 450-900 nm

Blue: 450-520 nm

Green: 520-600 nm

Red: 625-695 nm

NIR: 760-900 nm

Nadir:

Pan: 1m

MS: 4m

1-5 days, depending on latitude

Nadir:

8 km

50°

11 bits per pixel

Across track

and

along track

2003

5 -7 years

Active

FORMOSAT-2

Taiwan

Polar sun-synchronous

99.1 degrees

891 km

Pan: 450-900 nm

Blue: 450-520 nm

Green: 520-600 nm

Red: 630-690 nm

NIR: 760-900 nm

Nadir:

Pan: 2m

MS: 8m

daily Nadir:

24 km

45°

8 bits per pixel

Across track

and

along track

2004

5 years

Active

SPOT-4

2 HRV

Polar sun-synchronous

Pan: 610 - 680 nm

Nadir:

Pan: 10 m

2.5 days at 45° latitude

Nadir:

60 km

Twin HRV

8 bits per pixel

Across track 1998

5 years

74

France 98.8 degrees

820 km

MS:

Green: 500-590 nm

Red: 610-680 nm

NIR: 790-890 nm

SWIR: 1580-1750 nm

MS: 20 m configuration: 117 km

27°

Active

SPOT-5

2 HRG

France

Polar sun-synchronous

98.7 degrees

822 km

Pan: 480-710 nm

MS:

Green: 500-590 nm

Red: 610-680 nm

NIR: 780-890 nm

SWIR: 1580-1750 nm

Nadir:

Pan: 2.5m from 2 x 5m scenes

Pan: 5m

MS: 10m

SWIR: 20m

2 – 3 days, depending on latitude

Nadir:

60 km

Twin HRG configuration: 120 km

27°

8 bits per pixel

Across track

and

along track

2002

> 5 years

Active

IRS-1C/1D

India

Polar sun-synchronous

98.7 degrees

817 km

Pan: 500-750 nm

LISS-3:

Green: 520-590 nm

Red: 620-680 nm

NIR: 770-860 nm

SWIR: 1550-1700 nm

Pan: 5.8m

LISS-3:

VNIR: 23.5m

SWIR: 70m

Pan: 5 days

LISS-3: 24 days

Pan: 70km

LISS-3: 142km

Pan: 6 bits per pixel

LISS-3:

7 bits per pixel

Pan:

Across Track

IRS-1C 1995

IRS-1D 1996

3 years active

Not active

IRS-P5

(Cartosat-1)

India

Polar sun-synchronous

98.9 degrees

618 km

Pan: 500-850 nm

2.5 m 5 days 30 km 10 bit per pixel

Near simultaneous stereo images using 2 Pan cameras

2005

Nominal life time: 3 years

Active

IRS-P6 (Resourcesat-1)

India

Polar sun-synchronous

98.7 degrees

817 km

LISS-3

Green: 520-590 nm

Red: 620-680 nm

NIR: 770-860 nm

SWIR: 1550-1700 nm

LISS-4

MS Mode:

Green: 520-590 nm

Red: 620-680 nm

NIR: 770-860

LISS-3: 23,5m

LISS-4: 5,8m

AWIFS: 50 (Nadir) to 70m (off Nadir)

LISS-3: 5 days

LISS-4: 24 days (5 days 26° off nadir)

AWIFS: 5 days

LISS-3: 140km

LISS-4:

Pan: 70 km

MS: 23 km

AWIFS: 740 km

LISS-3:

7 bits per pixel

LISS-4:

10 bits per pixel (selected 7 bit are provided to the data handling system)

AWIFS:

10 bits per pixel

LISS-4:

Across Track

2003

Nominal life time: Until 2007

Active

75

nm

Pan Mode:

620-680 nm

AWIFS

Green: 520-590 nm

Red: 620-680 nm

NIR: 770-860 nm

SWIR: 1550-1700 nm

RapidEye

Constellation of 5 spacecrafts

Germany

Polar sun-synchronous

98 degrees

630 km

Blue: 440-510

Green: 520-590

Red: 630-685

Red edge: 690-730

NIR: 760-850

6.5 m Average repeat time at mid-latitude 5.5 days (Nadir)

Revisit time with body pointing

1 day

78 km (cumulative swath width of subsequent passages of all 5 satellites of the full constellation significantly larger)

25°

12 bit per pixel

Across track Spring 2008

Nominal life time: 7 years

TERRA

ASTER

USA

Polar sun-synchronous

98.2 degrees

705 km

VNIR:

Green: 520-600 nm

Red: 630-690 nm

NIR: 760-860 nm

SWIR:

1600-1700 nm

2145-2185 nm

2185-2225 nm

2235-2285 nm

2295-2365 nm

2360-2430 nm

TIR

8125-8475 nm

8475-8825 nm

8925-9275 nm

10250-10950 nm

VNIR: 15m

SWIR: 30m

TIR: 90m

16 days 60 km

Across track pointing:

VNIR: 318 km (24° off Nadir)

SWIR and TIR: 116 km (8.55° off Nadir)

VNIR: 8 bits per pixel

SWIR: 8 bits per pixel

TIR: 12 bits per pixel

Along Track 1999

Expected life time:

6 to 7 years

Active

76

10950-11650 nm

LANDSAT-7

ETM

USA

Polar sun-synchronous

98.2 degrees

705 km

Pan: 500-900 nm

MS:

Blue: 450-520 nm

Green: 530-610 nm

Red: 630-690 nm

NIR:780-900 nm

SWIR: 1550-1750 nm

SWIR: 2090-2350 nm

TIR: 10400-12500 nm

Nadir:

Pan: 15m

MS: 30m

TIR: 60m

16 days 185 km 8 bits per pixel

No 1999

7 years

Not properly working (data gaps)

Sources: http://www.satimagingcorp.com/satellite-sensors/ikonos.html http://www.sovzond.ru/en/satellites/indie/774.html http://ceos.cnes.fr:8100/cdrom-98/ceos1/isro/eospro/irs1c.htmhttp://www.isro.gov.in/pslvc5/index.html http://directory.eoportal.org/pres_IRSP6ResourceSat1.html http://www.digitalglobe.com/downloads/WV1_WV2_SpectralResponse.pdf

http://www.digitalglobe.com/about/worldview2.htmlhttp://directory.eoportal.org/pres_RapidEyeSatelliteConstellation.html http://directory.eoportal.org/pres_PleiadesHRHighResolutionOpticalImagingConstellationofCNES.html http://events.eoportal.org/pres_SPOT4.html

This wealth of remotely observed images related to construction needs to be converted into data (cf. Chapter 5.1.) which can be compared to official or private statistics dealing with the same sector. For commercial purposes a wide variety of software and data banks is available for this task. Companies offering these services are flourishing40.

To be more specific about what is entailed in converting remote sensing images in the area of housing construction, it is worth recalling a few details41. The surface area has to be characterized according to its use. Accuracy, resolution and periodicity of measurement have to be defined. Costs have to be assessed – and can range between € 3 and € 50/km2. The mode of conversion and the amount of attribute data have to be determined.

An empirical in-depth-analysis is then required to evaluate the behaviour of SOBI of housing in relation to conventional statistics. The major criteria for this task are:

40 For early overviews cf. Grant I. Thrall, GIS Applications in Real Estate and Related Industries, in: Journal of Housing Research, Vol. 9, Issue 1, 1998, or Luc Anselin, GIS Research Infrastructure for Spatial Analysis of Real Estate Markets, in: Journal of Housing Research, Vol. 9, Issue 1, 1998

41 Cf. Geoville loc.cit.

77

Does remote sensing enable the provision of time series of the housing SOBI so that they exhibit statistical paths over the housing cycles?

What are the specific time-patterns of the remote sensing data with the respective housing indicators?

What is the optimal and feasible geographical delimitation, considering this very detailed remote sensing data must be correlated with corresponding statistical indicators?

This exercise should lead to a much improved understanding of housing cycles and the acquisition of data far closer to real-time events, enhancing the value of housing as an indicator of business cycles in general.

7.2.1.2 Approaching the housing cycle via econometric models

At the high point of early econometric business cycle analysis42 in the 60s “investment in residential construction” was given importance, as it was a significant component of aggregate demand. Housing later played only a minor role in business cycle analysis. Measuring this component is mostly carried out using data from residential construction, hence from the supply angle. This is also the reason why equations explaining housing incorporate both demand and supply factors. During the early postwar decades residential and industrial construction investment seemed to follow a somewhat countercyclical pattern, the reason given being a lag with respect to fixed business investment. In contrast to this interpretation, more recent analysis favors a pro-cyclical shape of the housing function. Unfortunately, in up-to-date macroeconometric studies on business cycles, housing is only rarely integrated in the models themselves43. Exceptions exist44, and in particular mention has to be made of the thorough study by E. E. Leamer45 with the trenchant title: “Housing IS the business cycle”. Therein it is stated that:

Within the lists of BCI, indicators for housing or residential investment are scarce.

Theoretical explanations of housing as a function of income and interest rates are criticised as inadequate, since the explanatory power of this relationship is too shortsighted.

42 Cf. Michael K. Evans, Macroeconomic activity, New York, Evanston, London (Harper Int.) 1969, p. 184 ss.43 François Ortalo-Magné and Sven Rady , Housing transactions and macroeconomic fluctuations: a case study of England and Wales, in: Journal of Housing Economics Volume 13, Issue 4, December 2004, Pages 287-303purchase View Within Article

44 As to the data see the overview: D. Listokin, et al., Known Facts or reasonable Assumptions? An Examination of Alternative Sources of Housing Data, in: Journal of Housing Research, Vol. 13, Issue 2, Fannie Mae Foundation 2003, p. 219 ss.A recent econometric model is found in: M. Davis and J. Heathcote, Housing and the Business Cycle, Nov. 1, 2003, Finance Economics Discussion Paper No. 2004-11. SSRN: http://ssrn.com/abstract=528102

45 Edward E. Leamer, Housing IS the business cycle, NBER Working Paper 13428, www.nber.org/papers/w13428 , Cambridge, MA Sept. 2007

78

Empirically, it is evident that “Housing is the most important sector in our economic recessions”.

The importance of housing is elucidated as follows:

While residential investment contributes only a small fraction to total US GDP growth, its contribution to US recessions is huge. From the 50s to 2007, “Eight of the ten recessions were preceded by sustained and substantial problems in housing,”46. The reason given for the great importance of housing is that homes have a volume cycle, not a price cycle (because of price-stickiness).

Therefore the utmost attention should be paid to depicting housing cycles, not just for the general purpose of adequately describing business cycles but to ensure competent formulation of economic policy47. (The global crisis of 2008 is an unfortunate proof of the effects of failure in this area.)

There are specific requirements to focussing on housing as a prime area for SOBI.

The principles of the approach to be taken are straightforward:

Presentation and evaluation of suitable existing econometric models of housing and residential construction

Exploiting remote sensing data (SOBI) and integrating them into geoinformatics databanks

Expanding econometric models of housing with SOBI and geoinformatic data.

Initially, existing econometric models of the housing sector have to be evaluated with regard to their aptitude to integrate additional data derived from remote sensing sources. There is a great variety of such econometric models of housing and residential construction.48 The following model, developed by M. Davis and J. Heathcote49, is a good example. Traditionally, housing investment was regressed on house prices and costs of finance. Demand shocks were therefore the main explanation for residential investment behavior. In contrast, the Davis-Heathcote model is based on the concept of utility-maximization in a household production process within a multi-sectoral economy. It thus reflects the demand for intermediate goods (construction) derived from the 46 Loc.cit. p. 1347 Even the Great Depression seems to have been anticipated by a decline in housing beginning as early as 1925. “it

seems possible that the increase in the discount rate in 1928 was very hard on an already weakened housing sector, and set in motion the events that led to the Great Depression, dropping housing starts dramatically from over 900 thousand in 1925 to under 100 thousand in 1933.”

48 E.g. cf. M. I. Marshall and Th. L. Marsh, Consumer and investment demand for manufacturing housing units, Journal of Housing Economics, Vol. 16, Issue 1, March 2007, p. 59 -71, Akintola Akintoye ; Martin Skitmore , Models of UK private sector quarterly construction demand , in: Construction Management and Economics, Volume 12, Issue 1 January 1994 , p. 3 – 13, G. Cameron, J. Muellbauer and A. Murphy, Was there a British House Price bubble? Discussion Paper, Dep. of Economics, University of Oxford, Nr. 276, August 200649 M. Davis and J. Heathcote, Housing and the Business Cycle, Nov. 1, 2003, Finance economics Discussion Paper No.

2004-11. SSRN: http://ssrn.com/abstract=528102

79

demand of households for the consumer durable good which is “residential construction”. This approach allows assessment of the business cycle properties of residential investment with regard to other variables depicting the cycle and non-residential investment: “… residential investment is more than twice as volatile as business investment” and ”residential investment leads the business cycle whereas non-residential investments lags.”50 The model also provides the parameters for construction output, manufacturing output, etc.

As has been underlined in the survey of econometric business cycle models (chapter 4.2.) the intention of those who have developed the models is to explain and forecast the cycle with a minimum of variables and data. Yet, when it comes to disaggregating these models - which after all is the requirement of policy-makers at a national or regional level – elegance has to be sacrificed to arrive at the required detailed information. This means that highly aggregated housing models have to be brought down to earth – in both senses of the word. Thus, the geographical or regional dimension has to be introduced explicitly into the equations. This would, first, lead to a series of regional equations for residential housing which could look as follows:

ln(SOBIHSr,t ) = α+β ln(SOBIHSr, t-1)+ γ ln(Y/POPr,t-1)- δ ln(UCr,t-1)+z

where SOBIHS is the housing stock in region r at time t observed by satellite, Y/POP is the real disposable income per head, UC stands for the user costs of housing, i.e. mainly mortgage rates and rate of inflation and z represents other explanatory (shift) factors.

Depending on the overall econometric business cycle model it then remains to be determined if and how such single regional equations can be aggregated into the model. Spatial econometrics have to be applied which, in this case, implies that housing demand within the regions is interconnected to a certain degree. Housing prices and hence demand, for example, obviously depend not only on the prices in region but also on prices in other regions.

Thus, as with the indicator approach, the SOBI variables in the econometric approach have to be examined as regards their correlation with the respective conventional statistical indicators.

50 Loc.cit.

80

7.3. AGRICULTURE AND FORESTRY

From the perspective of business cycle analysis assisted by remote sensing, agriculture occupies an ambiguous place. On the one hand, there are few areas in which remote sensing has developed to such an advanced level worldwide as it has in agriculture and forestry. On the other hand, the weight of agriculture in highly industrialized countries in GDP is very small, ranging well below 5%. Little attention is therefore given to this sector in business cycle economics. Notwithstanding that, even with such a small proportion agriculture does deserve attention. Of course, for developing countries remote sensing of agriculture is of utmost importance.

Within the long lists of BCI, data on agriculture normally do not appear. However, data on agriculture abound where a longer-term perspective is adopted. Thus the OECD, for example, provides ample documentation on agricultural statistics. Its main economic indicators include the following detailed data on agriculture51:

“Total value of production (at farm gate)I. Total value of production (at farm gate) A. Share of Standard PSE (Producer Support Estimates) commodities

 II. Total value of consumption (at farm gate)II. Total value of consumption (at farm gate) A. Standard PSE commodities

 III. Producer Support Estimate (PSE)

III. Producer Support Estimate (PSE)

 A. Market price supportA. Market price support A.1. Standard PSE commodities

 B. Payments based on output

B. Payments based on outputB.1. Based on unlimited outputB.2. Based on limited output

 C. Payments based on area planted/animal numbersC. Payments based on area planted/animal numbers

C.1. Based on unlimited area or animal numbersC.2. Based on limited area or animal numbers

 D. Payments based on historical entitlements

D. Payments based on historical entitlements

D.1. Based on historical plantings/animal numbers or productionD.2. Based on historical support programmes

 E. Payments based on input use

E. Payments based on input useE.1. Based on use of variable inputsE.2. Based on use of on-farm servicesE.3. Based on use of fixed inputs

 F. Payments based on input constraintsF. Payments based on input constraints F.1. Based on constraints on variable inputs“

In the econometric modeling approach, agriculture can also look back on a long tradition. However, there is a qualification to be noted here, in that agriculture normally does not appear in overall business cycle models: “Very few macro-econometric models analyse the impact of agriculture, and other raw materials producers such as mining, in supply side terms through inter-sector relations

51 www.oecd.org/agr/support/psecse viz. OECD, Agricultural Policies in OECD countries, At a Glance 2008, Paris (OECD) 2008

81

with non-agrarian sectors even though this perspective is also important:”52 Mostly, agriculture is analyzed with regard to import/export relations and to personal (food) consumption expenditures. It is therefore clear that despite the direct small weight of agriculture in GDP its indirect and overall importance is reflected in its impact on demand components and interrelations with other industries.

The econometric supply analysis of agriculture can be illustrated with the following equation53:

IPRA stands for an Index of Price Ratios of Agriculture and QNA, of course, is non-agricultural production.

The nature of this calculation indicates that its primary focus is on determining trends. However, provided data are available, there is no impediment to switching to a cyclical analysis. Data for the dependent variable, i.e. agricultural production, could then be derived from remote sensing, ensuring timeliness of monitoring.

There is already a wealth of experience in such approaches using satellite observation. A few examples cited from the US may suffice:

52 M. C. Guisan and E. Pilar, Econometric models of agriculture in OECD countries, www.usc.es/economet/eaa.htm Univ. of Santiago de Compostela working paper No. 60, 2002

53 Guisan loc.cit.

82

Land Imaging54 as the overarching title

“The Landsat Program has changed the way we look at our planet. An entire new field for scientific study and practical applications had emerged: remote sensing. The Future of Operational Land Imaging Working Group is leading an effort to develop a long-term plan to achieve technical, financial, and managerial stability for operational land imaging in accord with the goals and objectives of the U.S. Integrated Earth Observation System.”

Within that program agriculture plays an important role, e.g.:

Agricultural production models55, by David E. Steitz, Headquarters, Washington

NASA and FAS

“NASA Satellites Improve Response To Global Agricultural Change:  NASA's Earth satellite observing systems are helping the U.S. Department of Agriculture Foreign Agricultural Service (FAS) improve the accuracy and timeliness of information they provide about important crops around the world. FAS information is crucial in decisions affecting U.S. agriculture, trade policy, and food aid.

NASA and the University of Maryland are providing the FAS with observations and data products from instruments on NASA's Aqua and Terra satellites and from the TOPEX/Poseidon, Jason and Tropical Rainfall Measuring Mission (TRMM) satellites. NASA provides daily, high-quality, observations of the Earth. The timeliness and quality of these science data products are used to support decision support tools employed by FAS to assess crop productivity over large areas of the world. NASA products allow FAS analysts to distinguish between different crops such as wheat and rice and permit analysts to measure other features like surface temperature and snow cover. Analysts can gauge the health of agriculture by comparing recent and historic data. NASA satellites collect data twice daily, Terra in the morning and Aqua in the afternoon. NASA's Rapid Response System processes and delivers observations to FAS usually less than four hours after it is collected. Scientists at the University of Maryland are creating an archive and an interface that enables analysts to compare current and historical conditions.”

Applications of NASA's Earth Science research enable the use of observations, measurements and models to improve agency partners' decision-making capabilities. FAS has benefited from incorporating products from Earth observation systems into operational procedures. “

56

GLAM—Global Agricultural Monitoring57

Project Background NASA/UDSA MOU54 http://www.landimaging.gov/55 http://www.pecad.fas.usda.gov/56 visit: http://www.gsfc.nasa.gov/topstory/2004/0115agriculture.html ,http://www.pecad.fas.usda.gov/cropexplorer/

57 http://www.pecad.fas.usda.gov/images/glam/GLAMFAS_brochure.pdf

83

The U.S. Department of Agriculture (USDA) and the National Aeronautics and Space Administration (NASA) recently signed a Memorandum of Understanding (MOU) to strengthen future collaboration. In support of this collaboration, NASA and the USDA Foreign Agricultural Service (FAS) jointly funded a new project to assimilate NASA's Moderate Resolution Imaging Spectroradiometer (MODIS) data and products into an existing decision support system (DSS) operated by the International Production Assessment Division (IPAD) of FAS. Building on NASA's investment in the MODIS Science Team, the project is implementing a user-friendly system that will allow for the integration and analysis of MODIS data products in IPAD's DSS.

The Application of NASA EOS MODIS Data to FAS Agricultural Assessment and ForecastingTo meet its objectives, FAS/IPAD uses satellite data and data products to monitor agriculture worldwide and to locate and keep track of natural disasters such as short and long term droughts, floods and persistent snow cover which impair agricultural productivity. FAS is the largest user of satellite imagery in the non-military sector of the U.S. government. For the last 20 years FAS has used a combination of Landsat and NOAA-AVHRR satellite data to monitor crop condition and report on episodic events.FAS is upgrading and enhancing the satellite component of its IPAD decision support system through an information delivery system for MODIS data and derived products. NASA's Moderate Resolution Imaging Spectroradiometer (MODIS) on board two platforms of the Earth Observing System (EOS), was designed in part to monitor subtle vegetation responses to stress, vegetation production and land cover with regional-to-global coverage. Hence, integration of MODIS data and derived products into the IPAD FAS DSS provides FAS with better characterization of land surface conditions at the regional scale and enables monitoring of changes in the key agricultural areas of FAS focus regions in a more timely fashion and at a higher resolution than previously possible with NOAA-AVHRR data.

Project Components:

Delivery and integration of MODIS Rapid Response data into the FAS monitoring system to facilitate improved monitoring of the impact of climate hazards, such as drought, large scale flooding, and snow storms, on agricultural production.

Development and delivery of a long term database of MODIS composite Vegetation Index (VI) time series including analysis tools and a graphic user interface that provides mosaicking, reprojection capabilities, and easy access to the moderate resolution image archive.

Establishment of the relationship between MODIS VI data and the long-term archives from the AVHRR and SPOT-VEGETATION used by FAS/IPAD.

Development of enhanced MODIS cropland products including a crop mask, a crop type map, new band combination products, and a crop stress index.

MODIS Rapid Response58

58 http://rapidfire.sci.gsfc.nasa.gov/

84

The U.S. Department of Agriculture's Foreign Agricultural Service (FAS) is using MODIS satellite data to predict crop yields. MODIS provides daily, high-quality, photo-like images that can be used to observe large areas across the world. These images help FAS improve the accuracy and timeliness of the crop yield predictions, which are needed to make decisions affecting U.S. agriculture, trade policy and food aid. MODIS products allow FAS analysts to distinguish between different crops like wheat and rice and permit analysts to measure other features like surface temperature and snow cover. Analysts can gauge the overall health of agriculture by comparing current data with previous years. MODIS products, first used by FAS in the summer of 2003, demonstrated their utility by helping analysts identify new areas of irrigated agriculture in the Middle East. On August 24, 2003, only the irrigated land was still green in the MODIS image on the left. MODIS collects data twice daily, from the Terra satellite in the morning and the Aqua satellite in the afternoon, which helps analysts observe how events such as fires, volcanic eruptions, floods, storms, or extreme temperatures affect crops. The MODIS Rapid Response System processes and delivers MODIS data to the USDA within three to four hours of it being acquired. FAS posts the data in the Crop Explorer. The University of Maryland also provides USDA crop analysts with a web interface for analyzing MODIS temporal composites of vegetation index data.”59

The above examples clearly demonstrate that the interface between satellite observation and agricultural models is firmly established. Hence, only a small step is involved in prioritising a short term perspective, so that remote observation of agriculture can be integrated into analysis of cyclical macroeconomic activity.

59 Global crop production estimates: http://www.pecad.fas.usda.gov/cropexplorer/

85

7.4 TRANSPORT

The role of transportation within business cycle analysis is disputed. On the one hand, transportation indicators do not seem to play a prominent role in business cycle indicator analysis and therefore are often omitted in business cycle models. On the other hand, transportation undoubtedly is a core mechanism in the economic machinery.

The importance of transportation is abundantly clear in the chart below60:

60 Lahiri, Kajal and Yao, Vincent, Economic Indicators for the US Transportation Sector. Transportation Research A: Policy and

Practice, SUNY at Albany, Vol. 40, No. 10, 2006, http://www.albany.edu/~klahiri/Lahiri_Yao_TR_A.pdf

86

With this in mind, it can be reasonably argued that, first, inventory cycles are an important element in fluctuations of GDP and, second, transportation is closely related to these inventory cycles. Beyond these links, the interrelations of transportation with the aggregate economy are complex:

Applying the BCI-approach:

In order to capture this complex interplay, considerable care has to be taken in selecting transport indicators which reflect these links61:

“

Here, the basic question of this study again presents itself: if and to what extent SOBI can deliver better information on transportation. It appears that the role of SOBI would relate to very specific queries. This is because monitoring of transportation, volumes, data and statistics is already well-developed, be it for railways, roads, ships or aircraft62. Thus satellite data would serve mostly as a corollary and supplement. They could serve, for instance, a useful purpose where there are

61 Lahiri loc.cit.62 Cf. e.g. OECD, Transport 2008, vol. 2008, no. 5,Trends in the Transport Sector: 1970-2006 , Paris, 2008 Edition

87

questions to be raised about the objectivity of “official” statistics. There are known cases in which statistics have obviously been manipulated for lobbying purposes. SOBI could play a role in monitoring and correction. Such data offers a very exact picture of the number and speeds of vehicles on the move. Associated with attribute data on the location of manufacturing plants, this would also enable estimates of the degree of activity of these specific industries and locations. Thus combining remote sensing with conventional indicators of transport has significant advantages.

Applying the econometric approach:

Analyzing transport with the help of econometric models is also well developed. Naturally, models differentiate between the kinds of transportation. Railways may serve as an example63:

The following graph illustrates the model:

63 P.S. Rao, Forecasting the Demand for Railway Freight Services, in: Journal of Transport Economics and Policy, Vol. 12, No 1, January 1978

88

The chart reveals quite clearly that demand for freight services depends on the output of factories, disaggregated by industries. This confirms the contention that appropriately selected transportation indicators reflect the “steam” of (cyclical) economic activity. In this vein, the model constructed on the above chart has to be hybrid, in the sense that very detailed information on the industries has to be incorporated by input/output structures64. A glance at the model65 quickly reveals that data to empirically estimate such a model can be fitted in both from “ordinary” statistical sources or from SOBI:

64 Rao:

65 Rao, loc.cit p. 9 s.

89

The role of remote sensing data in enhancing similar models for road or other forms of transportation may also be investigated. Again, the key benefit of including SOBI on transportation seems to be to deliver near-real-time objective data.

90

7.5. PRODUCTION CAPACITY AND EMISSIONS

Producers react most quickly to changes in demand by modifying their production capacity. This is well ahead of adapting production via additional investment (machinery) or manpower. Measures of capacity are manifold. At the simplest level, it is important to distinguish between technical and economic measures of capacity. Because of the sensitivity of changes in capacity, economists are most interested in gaining insight into “true” capacity indicators. As data on capacity utilization is often inadequate, it is common practice to resort to artificial calculations of production activity. The divergence between measured GDP and potential GDP (or output) is the classical macroeconomic example. Its importance derives from the fact that macroeconomic policy aims to close the “output gap” by utilising the arsenal of monetary and fiscal instruments66. Potential output is defined by the Congressional Budget Office as follows67:

“For the Congressional Budget Office (CBO), estimating the potential output of the economy and projecting future levels of that output are integral parts of producing short-term economic forecasts and medium-term economic projections. Potential output is an estimate of "full-employment" gross domestic product, or the level of GDP attainable when the economy is operating at a high rate of resource use. Rather than being a technical ceiling on production, potential GDP is a measure of the economy's maximum sustainable output, in which the intensity of resource use is neither adding to nor subtracting from inflationary pressure. There are many ways to compute the economy's productive potential. Some methods rely on purely statistical techniques. Others--including CBO's method--rely on statistical procedures grounded in economic theory.”

While this definition relies predominantly on theoretical macroeconomics, for the present purpose a more empirical understanding is needed. This can be achieved by using empirical measures of capacity utilization. Fluctuations of this measure serve as the prime indicator of business cycles68. The Institute for Economic Analysis (IEA) also follows this line of thought69: “Operating Rate of the Economy (ORE) or Potential Utilization Rate (PUR)-- actual output as percentage of the trend rate of "potential" GDP. As an empirical "indicator" this macroeconomic measure is conceptually similar to the familiar "capacity utilization rate" of manufacturing industry (BEA indicator #82, p.C-14).”

66 Cf. EU, Economic Policy Committee, Report on Potential Output and the Output Gap, Brussels, 25 October 2001,ECFIN/EPC/670/01/EN: :

67 http://www.cbo.gov/doc.cfm?index=5191&type=068 In this sense cf. the German ifo-Institute: „In der modernen Konjunkturtheorie sind Konjunkturzyklen als Schwankungen des Auslastungsgrades des gesamtwirtschaftlichen Produktionspotentials definiert. Originare vierteljahrliche Daten uber

die Kapazitatsauslastung im verarbeitenden Gewerbe und im Baugewerbe werden in Deutschland vom ifo Institut im Rahmen des ifo Konjunkturtests (KT) veroffentlicht.“69 http://www.iea-macro-economics.org/leading.html

91

An examination of actual tables on capacity utilization is informative here70:

Release Date: October 16, 2008

The importance of capacity utilization is also exemplified by the following71: “Revision of Industrial Production and Capacity Utilization: The Federal Reserve Board plans to issue its annual revision to the index of industrial production (IP) and the related measures of capacity utilization in late March of 2009. The revised IP indexes will incorporate data from selected editions of the U.S. Census Bureau's 2007 Current Industrial Reports. Detailed data from the 2007 Economic Census, however, are not expected to be available. Annual data from the U.S. Geological Survey regarding metallic and nonmetallic minerals (except fuels) for 2007 will also be incorporated. The updating will include revisions to the monthly indicator (either product 70 http://www.federalreserve.gov/releases/g17/Current/table7.htm71 http://www.federalreserve.gov/releases/g17/Current/default.htm

92

data or input data) and to seasonal factors for each industry as well as changes in the estimation methods for some series. Any changes to the methods for estimating the output of an industry will affect the index from 1972 to the present.”

It is clear from the above discussion and examples that estimates play a considerable role in data on capacity utilization. In this regard, greater exactitude should be welcome. Can remote sensing contribute to gathering more exact information on capacity utilization?

A few detours are unavoidable here:

The crucial point is to identify a technical marker which indicates a change in the capacity utilization of a factory. Technology differs across industries. A pattern would have to be identified which would indicate where and to what extent changes in capacity utilization are associated with different kinds of emissions. In the absence of such patterns – which, moreover, change over time – it can only be speculated that changing capacity would be reflected in thermal indicators, such as pollution, emissions of nitrogen dioxide NO2 etc. If these assumptions are accurate, they suggest a wide spectrum in which remote observation could usefully operate. Much experience has already been gained in gathering data on earth warming, pollution and the like. A few examples may suffice:

Tracking air pollution is the task performed by TEMIS (Tropospheric Emission Monitoring Internet Service) of ESA72. This service provides near real-time data on NO2, ozone, sulfur dioxide and other pollutants. Also, thermal radiation is a standard measure in remote sensing73. In general, achievements in tackling the problem of global warming74 can also be of use – on a much more modest scale - for assessing the changing rate of capacity utilization of factories.

72 Cf. Geoville, loc.cit. and http://www.temis.nl/

73 E.g. C. P. LO; D. A. QUATTROCHI; J. C. LUVALL, Application of high-resolution thermal infrared remote sensing and GIS to assess the urban heat island effect, in: International Journal of Remote Sensing, Volume 18, Issue 2 January 1997 , pages 287 - 304

74 http://earthobservatory.nasa.gov/Library/GlobalWarmingUpdate/global_warming_2002.pdf

93

Clearly, there remains much to be done to establish reliable “hybrid indicators” which include both satellite observed data and terrestrial data to provide information on capacity utilization changes in manufacturing. Large lacunae remain in remote sensing and economic models in this area.

94

8 CONCLUSIONS AND OUTLOOK

The aim of this study has been to demonstrate that observation from satellites is able to speed up monitoring and analysis of business cycles. Timeliness ranks amongst the top priorities for improving business cycle analysis. As has been demonstrated, publication of the latest business cycle indicators may trigger reactions on the stock market, in consumption and investment behavior and, perhaps, in decision-making in the areas of fiscal and monetary policy. Any reduction in the time-lag between real events and the recognition of these events in the respective business cycle analysis may improve individual behavior and public policy-making. Conventional methods to improve the quality of data have reached their limits. It is difficult to see where improvements could come in speeding up the gathering and processing of business cycle statistics via questionnaires.

Ample evidence has been presented here on how and where remote sensing could play a role. However, it cannot be assumed that economists will embrace remote sensing without hestitation. Both the potential areas of impact and the technical procedures involved are perhaps not self-evident. It is hoped, however, that the examples and procedures presented in the study stimulate further engagement with this topic.

In particular, it should be noted that the obvious potential of remote sensing is only a step away from actual and extended application in those areas described in chapter 6. Tapping this potential would involve techniques already familiar from the testing of econometric models. Once the statistical database derived from SOBI and attribute data is established, the performance of this combined approach has to be tested. This is usually done by “ex post forecasts”. In practice, known figures are “forecast” based on a hybrid model of satellite and terrestrial data.

In conclusion, there is good reason to assert that remote sensing offers great potential and a vital new approach to improving business cycle analysis. The challenge is to bold minds in both areas, remote sensing and economics, to tap that potential and bring about further advances.

95

References

Hagedh Bouakez, Emanuela Cardia, Francisco J. Ruge-Murcia (2005): The Transmission of Monetary Policy in a Multi-Sector Economy, CIREQ, Montreal, Mimeo.

Runar Brännlund, Tarek Ghalwash, and Jonas Nordström (2007) Increased Energy Intensity and the Rebound Effect: Effects on Consumption and Emissions, Energy Economics, 29 (1),

Flint Brayton, Anrdew Levin, Ralph Tyron, and John C. Williams (1997): The Evolution of Macro Models at the Federal Reserve Board, Federal Reserve Board, Washington, D.C..

Bureau of Economics Analysis, 1984. Handbook of Cyclical Indicators. US Department of Commerce, Washington, DC.

Burns, A.F., Mitchell, W.C., 1946. Measuring Business Cycles- National Bureau of Economic Research, New York.

Arthur F. Burns and Wesley C. Mitchell, Measuring Business Cycles, 1946, p.2

M. Chauvet and J. Piger, A Comparison of the Real-Time Performance of Business Cycle Dating Methods, Working Paper, University of Oregon, Feb 9, 2007, cf. also: M.W. Watson, Business Cycle Duration and Postwar Stabilization of the U.S. Economy, in: American Economic Review, 84, 1994, p. 24-46, or: Journal of Business Cycle Measurement and Analysis: Volume 3, No. 1, OECD, 2007

The Conference Board, Business Cycle Indicators Handbook

The Conference Board, Business Cycle Indicators, monthly reports

Cooperation of member institutes is assured by AQUIEC, the Association for the Quality of the Economic Indicators of the Construction Industry

96

M. Davis and J. Heathcote, Housing and the Business Cycle, Nov. 1, 2003, Finance economics Discussion Paper No. 2004-11. SSRN: http://ssrn.com/abstract=528102

Thomas Dalsgaard, Christophe André, Pete Richardson (2001): Standard Shocks in the OECD Interlink Model, OECD Economics Department Working Papers, No. 306, OECD Publishing.

Francis X. Diebold (1998): The Past, Present, and Future of Macroeconomic Forecasting, Journal of Economic Perspectives, 12, 175-192.

Martin Distelkamp, Frank Hohmann, Christian Lutz, Bernd Meyer, and Marc Ingo Wolter (2003): Das IAB/INFORGE Modell – Ein neuer ökonometrischer Ansatz gesamtwirtschaftlicher und länderspezifischer Szenarien, Beiträge zur Arbeitsmarkt- und Berufsforschung, 275, Institut für Arbeitsmarkt und Berufsforschung.

James S. Duesenberry, Gary Fromm, Lawrence W. Klein and Edwin Kuh, Eds., The Brookings Quarterly Econometric Model of the United States, Rand McNally, 1965

Cf. EU, Economic Policy Committee, Report on Potential Output and the Output Gap, Brussels, 25 October 2001, ECFIN/EPC/670/01/EN: :

Cf. Michael K. Evans, Macroeconomic activity, New York, Evanston, London (Harper Int.) 1969, p. 184 ss.

Fernerkundung in der Konjunkturforschung, GeoVille Information Systems GmbH, Bericht an FFG, Innsbruck, 12 01 2008

Robert Fildes, Herman Stekler (2002): The State of Macroeconomice Forecasting, Journal of Macroeconomics 24 (2002) 435-468.

Cf.e.g. L. Van Genderen, C. Pohl J.Review article Multisensor image fusion in remote sensing: concepts, methods and applications, in: International Journal of Remote Sensing, Volume 19, Number 5, 1998 , pp. 823-854(32)

97

Global crop production estimates: http://www.pecad.fas.usda.gov/cropexplorer/

M. C. Guisan and E. Pilar, Econometric models of agriculture in OECD countries, www.usc.es/economet/eaa.htm Univ. of Santiago de Compostela working paper No. 60, 2002

Andreas Hornstein, Jack Praschnik (1994): The Real Business Cycle: Intermediate Inputs and Sectoral Comovement, Institute for Empirical Macroeconomics, Federal Reserve Bank of Minneapolis, Discussion Paper 89.

In: Journal of Business Cycle Measurement and Analysis 2003, vol. 2004, no. 1, pp. 116 – 141

In this sense cf. the German ifo-Institute: „In der modernen Konjunkturtheorie sind Konjunkturzyklen als Schwankungen des Auslastungsgrades des gesamtwirtschaftlichen Produktionspotentials definiert. Originare vierteljahrliche Daten uber die Kapazitatsauslastung im verarbeitenden Gewerbe und im Baugewerbe werden in Deutschland vom ifo Institut im Rahmen des ifo Konjunkturtests (KT) veroffentlicht.“

Journal of Business Cycle Measurement and Analysis 2007, vol.2007, no.1, pp.103-125; Report: Improving Timeliness for Short-Term Economic Statistics; cf. www.oecd.org/std/research/timeliness

Kurt Kratena, Ina Meyer, and Michael Würger (2008): Modelling the Energy Demand of Households in a Combined Top Down/Bottom Up Approach, Austrian Institute of Economic Research (WIFO), Working Papers (321/2008).

Lahiri, Kajal and Yao, Vincent, Economic Indicators for the US Transportation Sector. Transportation Research A: Policy and Practice, SUNY at Albany, Vol. 40, No. 10, 2006, http://www.albany.edu/~klahiri/Lahiri_Yao_TR_A.pdf

Edward E. Leamer, Housing IS the business cycle, NBER Working Paper 13428, www.nber.org/papers/w13428, Cambridge, MA Sept. 2007

De Leeuw, F., 1991. Toward a theory leading indicators. In: Lahiri, K., Moore, G.H. (Eds.), Leading Economic Indicators: New Approaches and Forecasting Records. Cambridge University Press, pp. 15-56.

98

Yvan Lengwiler (2004): Microfoundations of Financial Economics – An Introduction to General Equilibrium Asset Pricing, Princeton Series in Finance, Princeton University Press.

As to the data see the overview: D. Listokin, et al., Known Facts or reasonable Assumptions? An Examination of Alternative Sources of Housing Data, in: Journal of Housing Research, Vol. 13, Issue 2, Fannie Mae Foundation 2003, p. 219 ss. A recent econometric model is found in: M. Davis and J. Heathcote, Housing and the Business Cycle, Nov. 1, 2003, Finance Economics Discussion Paper No. 2004-11. SSRN: http://ssrn.com/abstract=528102

Christian Lutz, Martin Distelkamp, Bernd Meyer, and Marc Ingo Wolter (2003): Forecasting the Interindustry Development of the German Economy: The Model INFORGE, GWS Discussion Paper 2003/2, Gesellschaft für Wirtschaftliche Strukturforschung mbH (GWS).

E.g. cf. M. I. Marshall and Th. L. Marsh, Consumer and investment demand for manufacturing housing units, Journal of Housing Economics, Vol. 16, Issue 1, March 2007, p. 59 -71, Akintola Akintoye ; Martin Skitmore , Models of UK private sector quarterly construction demand , in: Construction Management and Economics, Volume 12, Issue 1 January 1994 , p. 3 – 13, G. Cameron, J. Muellbauer and A. Murphy, Was there a British House Price bubble? Discussion Paper, Dep. of Economics, University of Oxford, Nr. 276, August 2006

Moore, G.H., 1961. Business Cycle Indicators, vol. 1. Princeton University Press for NBER, Princeton, NJ.

L. Rachel Ngai, Christopher A. Pissarides (2005): Structural Change in a Multisector Model of Growth, London School of Economics, Mimeo.

Konstantinos Nikolopoulos, Robert Fildes, Paul Goodwin, and Micheal Lawrence (2005): On the Accuracy of Judgmental Interventions on Forecasting Support Systems, Lancaster University Management School Working Paper 2005/022.

Cf. E.g. Viggo Nordvik, Selective housing policy in local housing markets and the supply of housing , In: Journal of Housing Economics, Volume 15, Issue 4, December 2006, p. 279 ss.

Cf. e.g. OECD, Transport 2008, vol. 2008, no. 5,Trends in the Transport Sector: 1970-2006 , Paris, 2008 Edition

Maureen O’Hara (1995): Market Microstructure Theory, Blackwell Business, Cambridge, Massachusetts.

99

One outstanding example of BCI-research is Geoffrey H. Moore, Ed. Business Cycle Indicators, NBER, Princeton 1961 within the realm of the National Bureau Of Economic Research (NBER)

François Ortalo-Magné and Sven Rady , Housing transactions and macroeconomic fluctuations: a case study of England and Wales, in: Journal of Housing Economics Volume 13, Issue 4, December 2004, Pages 287-303

Nigel Pain, Annabelle Mourougane, Franck Sédillot, and Laurence Le Fouler (2005): The New OECD Trade Model, OECD Economics Department Working Papers, No. 440.

P.S. Rao, Forecasting the Demand for Railway Freight Services, in: Journal of TransportEconomics and Policy, Vol. 12, No 1, January 1978

Marco Ratto, Werner Röger, Jan in’t Veld, and Riccardo Girardi (2005): An estimated New Keynesian Dynamic Stochastic General Equilibrium Model of the Euro Area, Economic Papers No. 220, Directorate-General for Economic and Financial Affairs, European Commission.

Pete Richardson (1988): The Structure and Simulation Properties of OECD’s Interlink Model, OECD Economic Studies, No. 10.

Werner Roeger, Jan in’t Veld (1997),: QUEST II: A Multi Country Business Cycle and Growth Model, Economic Papers No. 123, European Commission.

Werner Röger, Jan in’t Veld (2002),: Some Selected Simulation Experiments with the European Commission’s QUEST Model, Economic Papers No. 178, Directorate-General for Economic and Financial Affairs, European Commission.

Simon Fraser University Library

wien 2001

100

Karl W. Steininger, Birgit Friedl, Brigitte Gebetsroither (2006): Sustainability Impacts of Car Road Pricing: A Computational General Equilibrium Analysis for Austria, Ecological Economics (2006).

G. I. Thrall, GIS Applications in Real Estate and Related Industries, in: Journal of Housing Research, Vol. 9 Iss. 1, 1998 p. 65 ss., L. Anselin, GIS Research Infrastructure for Spatial Analysis of Real Estate Markets, in: Journal loc. cit. p. 77 ss.

For early overviews cf. Grant I. Thrall, GIS Applications in Real Estate and Related Industries, in: Journal of Housing Research, Vol. 9, Issue 1, 1998, or Luc Anselin, GIS Research Infrastructure for Spatial Analysis of Real Estate Markets, in: Journal of Housing Research, Vol. 9, Issue 1, 1998

University of New Hampshire (2006, November 20). Satellite Observation Tracks Avian Flu. ScienceDaily. Retrieved July 22, 2008, from http://www.sciencedaily.com¬/ releases/2006/11/061120101546.htm

Alpo Willmann, Mika Kortelainen, Hanna-Leena Männistö, and Mika Tujula (1998): The BOF5 Macroeconomic Model of Finland, Structure and Equations, Bank of Finland Discussionpapers.

www.oecd.org/agr/support/psecse viz. OECD, Agricultural Policies in OECD countries, At a Glance 2008, Paris (OECD) 2008

Zarnowitz, V., 1992. Business Cycles: Theory, History, Indicators, Forecasting. The University of Chicago Press, Chicago.

Zarnowitz, V., Boschan, C., 1977. Cyclical Indicators: An Evaluation and New Leading Indexes. US Department of Commerce, pp. 170-179.

Websites:

http://envisat.esa.int/

http://modis.gsfc.nasa.gov/

http://nasadaacs.eos.nasa.gov/articles/2007_menu.html

http://nasadaacs.eos.nasa.gov/index.html

101

http://stats.oecd.org/wbos/Index.aspx?querytype=view&queryname=93

http://rapidfire.sci.gsfc.nasa.gov/

http://rapidfire.sci.gsfc.nasa.gov/apps/

http://www.cbo.gov/doc.cfm?index=5191&type=0

http://www.census.gov/epcd/econ/www/indijun.htm

http://www.census.gov/const/www/nrcdatarelationships.html

http://www.conference-board.org/economics/bci/

http://www.conference-board.org/pdf_free/economics/bci/BCI-Handbook.pdf

http://www.esri.com/software/arcgis/arcinfo/about/features.html

http://www.esri.com/software/arcgis/geodatabase/index.html

http://www.euroconstruct.org/ EUROCONSTRUCT provides detailed up-to-date construction market forecasts for all main construction sectors: residential construction, non-residential construction and civil engineering.

http://www.euroconstruct.org/publications/cr_sample.pdf

http://www.federalreserve.gov/releases/g17/Current/default.htm

http://www.federalreserve.gov/releases/g17/Current/table7.htm

http://www.geospatial-solutions.com/geospatialsolutions

http://www.gsfc.nasa.gov/topstory/2004/0115agriculture.html

http://www.iea-macro-economics.org/leading.html

http://www.investopedia.com/university/releases/housingstarts.asp

http://www.landimaging.gov/

http://www.pecad.fas.usda.gov/images/glam/GLAMFAS_brochure.pdf

http://www.pecad.fas.usda.gov/cropexplorer/

http://www.pecad.fas.usda.gov/

http://www.safe.com/products/desktop/overview.php

http://www.temis.nl/

www.gis.com

www.stern.nyu.edu/~nroubini/bci/bci.html

102

103

ANNEX:

AN OVERVIEW OF BUSINESS CYCLE INDICATORS

The Economic Indicators Program of U.S. Census Bureau is downloadable on: http://www.census.gov/cgi-bin/briefroom/BriefRm

And includes:

The Census Economic Briefing Rooms page contains links to all indicator reports. There are also release schedules for the reports on this page.  Our indicators include:

Construction spending Durable goods, new orders Homeownership Housing starts Housing vacancy Manufacturers' new orders Manufacturers' profits Monthly retail sales   Monthly wholesale trade New home sales Quarterly services survey   Retail profits   Total business sales Trade balance OECD’s Main Indicators

INCLUDEPICTURE "http://stats.oecd.org/mei/TitleMei.gif" \* MERGEFORMATINET

OECD Business Cycle Analysis DatabaseSearch OECD Statistics

  INCLUDEPICTURE "http://stats.oecd.org/mei/picto1.gif" \* MERGEFORMATINET INCLUDEPICTURE "http://stats.oecd.org/mei/picto1.gif" \* MERGEFORMATINET   Sources, Definitions & Public Data (Home)    Real-Time & Revisions Database

  This interface provides access to time series data for OECD Composite Leading Indicators (CLI), standardised consumer and business confidence indicators, business tendency survey indicators by sector and consumer opinion survey indicators as published in each monthly edition of the OECD Main Economic Indicators. Accessing this source data will enable users to analyse in depth the development of business cycles within and across countries. Data for all OECD Member countries (except Iceland), the Euro area, China, India, Indonesia, Brazil, South

104

    Business Cycle   Analysis   Databas e    Unit Labour   Cost   Indicators     Key Economic          Indicators   INCLUDEPICTURE "http://stats.oecd.org/mei/picto1.gif" \* MERGEFORMATINET INCLUDEPICTURE "http://stats.oecd.org/mei/picto1.gif" \* MERGEFORMATINET   Buy Full Database or   Publication     MEI Home    Statistical Sources    Contact Us

Africa and the Russian Federation are available.

NEWS & LATEST INFORMATION

Latest CLI press release

Revision Analysis of CLI (2007)

CLI for Major Emerging Economies (2007)

DATA & METADATA

OECD Composite Leading Indicators

Database Country

Component Series & Reference Turning Point Dates

Geographic Zone Weights

Terminology

Business Tendency and Consumer Opinion Surveys

OECD Standardised Confidence Indicators

Business Tendency Survey Database

Consumer Opinion Survey Database

Terminology

METHODOLOGY & ANALYSIS

OECD Composite Leading Indicators

OECD CLI - a Tool for Short-Term Analysis (2001)

CLI for OECD Non-Member Economies (2006)

Review of OECD CLI (2002)

External Evaluation of Euro Area Leading Indicators (2001)

Euro Area CLI - Comparison of Methods (2001)

OECD CLI - Meeting User Needs (2000)

Compiling CLI - A Comparison of Two Methods (2000)

The 1994 Mexican Crisis - Signals from Leading Indicators (1997)

Business Tendency and Consumer Opinion Surveys

International Development Work & Coordination

OECD / CIRET Journal of Business Cycle Measurement and Analysis

OECD Standardised Confidence Indicators (2005)

105

Business Tendency Surveys: A Handbook (2003)

Country specific vs Harmonised Confidence Indicators (2003)

Harmonisation Between EU and non-EU Countries (2003)

Consumer Surveys: Methodology & Analytical Use (2000)

Confidence Indicators & Composite Indicators (2000)

Cyclical Indicators & Business Tendency Surveys (1997)

 Variables included in the databases

Data is available for the variables listed below. Some variables listed may not be available for all countries.

OECD Composite Leading Indicators Database:

Composite Leading Indicators: amplitude adjusted, trend restored, 6-month rate of change

CLI Reference series (Industrial production): original, trend, ratio to trend

OECD Standardised Confidence Indicators: Consumer Confidence, Business Confidence (Manufacturing sector)

Business Tendency Survey Indicators Database:

Manufacturing industry: Production, Finished goods stocks, Raw materials stocks, Order books, Order inflow, Total demand, Export orders, Selling prices,

106

Employment, Capacity utilization, Business situation, Confidence indicator

Construction industry: Order books, Order inflow, Selling prices, Employment, Business situation, Confidence indicator

Retail trade: Order intentions / demand, Volume of stocks, Employment, Business situation, Confidence indicator

Service industries: Demand evolution, Employment, Business situation, Confidence indicator

Consumer Opinion Survey Indicators Database:

Confidence indicator, Expected inflation, Expected economic situation

Ifo Institute

107

108

109