In an economic class I took this summer, I had a chance to watch a presentation and paper given by Dean Baker, professor of economics at Bucknell University. The topic he presented is about recent financial meltdown due to housing market bubble and subprime morgage.
As far as I understand from his presentation, the Federal reserve or the economic policy makers such as Allan Greenspan failed to pay attention to the data which suggests the rising price and demand in housing marketing is nothing but a bubble. Couple of reasons are as follow.
- Dispairty between housing prices and income – although housing prices had been rising by double over the last recent few years, the income of American people did not go in parallel.
- Low demand for renting – over the same period during housing market bubble, the demand for rent is low comparing to demand for buying or selling houses. So, vaccancy in houses was rising across the nation.
He also related to housing bubble with stock market bubble which happened priror to housing bubble. And also wrong incentives among business people in the housing market. For example, rewarding people who can make more morgage deals in terms of quantity regardless of the quality of the loans, and inflated judgement by appriasers to favor the morgage broker.
Anyhow, above data findings are warning and signs of something going wrong with the housing market. So, the economic policy makers paid attention such warning signs and they could have prevented by issuing warning to the public, raising interest rates and imposing regulations to change wrong incentives.
So in this case, I believe that Information Technology(IT) can help policy makers in giving alerts and a better understanding of the state of economy. What if there is data engine that can provide reports and alerts of the state of the economy using mathematical models, based data collected from various sources across the country? Such an engine can be built using three layers of abstractions, as follow in the diagram below.
Economic Data Engine
1. Data Input layer – this layer is to collect data from different sources. This can be difficult and tendious job. In computer science, one of the biggest challenges is getting data from various sources and data formats. For example, person A might record data in PDF format, and person B record data in MS Word, or JPEG etc. And also the data structure can be different too. For solving such issues, we can have multiple data feed adapters which are built to consume data specifically from particular sources or formats. When there is new data format or source coming in, developers can create new data feed adapter for that source and plug it in to the system.
2. Model layer – This layer takes the output of data layer and tries to digest the data into a meaningful information using mathematical models. Mathematicians, statisticians and researchers can develop several models reflecting to the changes in the economy ocassionally. The system should be able to use one or more models at the same time to give different points of view of information.
3. Report View layer – This layer will extract the information provided by model layer into a human readable or user-friendly formats. Policy makers can use this layer to search for data, view state of the economy, and receive early warning signs of upcoming economic bubble etc.. and get the understanding of where the economy is heading. Tech companies like Google can contribute a lot in search technology and even in data consuming of the first layer.
Data layer and report view layer will be “sticky”. By which I mean data source formats are unlikely to change much in the short-run once data feed adapters are created. And also policy makers will not expect much change in the way they view the data reports in short-run. However, model layer should be volatile and agile. Because models are usually simplified version of the reality while it misses to take into other factors or notice the changing factors. Therefore, new models should created or modifying the existing ones to reflect the change in the nature of the ecomony.
There will be a lot of barriers to implement such data engine in a technologically advanced country like United States. The task to digitize information across the whole spectrum of economy in the country can be too costly and overwhelming. One way to reduce cost can be cooperation of nations to develop such a data engine and share the costs. Once it is built, the data engine should be able to use anywhere in the world with some changes to fit the different natures of economies in their specific countries. Other difficulties can include authenticity of data coming in when there is corruption in a country.
After all, such data engine is just a tool and if policy makers do not make use of it, it will be just like a piece of paper on the table. I am just imagining and so such ecnomic data engine might not be practically feasible. Let me know what you think.
FYI: the link to Dean Baker’s presention is here : http://www.vimeo.com/5578978 .