Multivariate Time Series Modeling
as a means of Rhyming the Reason to Harbor Deepening for Post-Panamax Shipping

Share This:



Multivariate Time Series Modeling

Herbert M Barber, Jr, PhD, PhD
Xicon Economics

The case for the PORT of JACKSONVILLE as presented by the economist who conducted the financial and economic argument for harbor deepening at JAXPORT, Dr. Herbert Barber.

The US maritime industry is under definitive transition as ports scramble to deepen harbors in satisfying container and vessel requirements of inbound and outbound post-Panamax ships. As anticipated, such deepening has not been without immense controversy, controversy where the polarity could not be greater.

As with many publicly funded infrastructure projects, deepening of the St. Johns River for the Port of Jacksonville will be an arduous endeavor with many potential landmines that must be navigated as the six-year deepening project is completed. In consideration of harbor deepening, government officials, politicians, and planners have an arsenal of tools at their disposal if they only choose to consider them. However, these tools are not in the toolbox of engineers, but rather, economists. Unfortunately, most entities often put the cart before the horse, beginning the planning phase with conceptual engineering design while negating sound economic analyses altogether; or by using economic analyses and projections rendered by engineers that are without scientific merit. For example, in consideration of a new port along the East Coast, engineers recently rendered cargo projections for 50 years. Fifty years of projections with no historical data upon which to draw. Wow! Even a novice (PhD level) researcher realizes the err of these projections, not to mention the ridiculousness. Without understanding the foolishness of using such ridiculous projections, the client is likely to move forward with the multi-billion dollar endeavor with no definitive cargo projections.

Unfortunately, this situation is more common than not—the rule rather than the exception. Decision-makers regularly use weak analyses and projections to render decisions regarding the spending of public monies for large infrastructure endeavors as a means of increasing economic output. The literature, both professional and scientific, contains a graveyard of large public spending projects gone awry.

While most politicians seem to favor infrastructure spending, as their next election is always pending; and most engineers favor infrastructure spending, as their next contract is right around the corner, the decision to enter large publically funded infrastructure endeavors is not for the faint-hearted. The decision is in fact, quite complex, with numerous variables operating simultaneously for and against the decision. Regardless, stakeholders are left to sift through a maze of complex economic and financial studies, not to mention a plethora of engineering and environmental studies.

To this end, researchers have numerous quantitative options available to them to determine project feasibility, not the least of which is multivariate time series analysis. Such was the case in anticipation of deepening the St. Johns River for JAXPORT in preparation of handling post-Panamax ships. As the engineering economics consultant for the deepening project, we eventually settled on multivariate time series as our primary forecasting technique for several variables, particularly those associated with inbound and outbound commodities. In so doing, of course, internal debates were made for and against various analytical techniques for analyzing and forecasting commodities at JAXPORT.

Multivariate time series analysis allows researchers to render statistically robust projections; so it was with commodities at JAXPORT. In so doing, time series allowed us to manage seasonality, trend, and irregularity under a single technique. For JAXPORT, we opted for a moving average technique to smooth data rather than other techniques, such as exponential smoothing.

Of course, as the name suggests, time series presumes that some form of natural temporal order exists within each series. This in and of itself is sometimes difficult to access when dealing with cargo data, as inbound and outbound commodities can be sporadic, lending itself to no observable natural order, altogether. In such a case, cross-sectional analyses may better serve as an analytical technique. Conversely to time series analyses, cross-sectional analyses allow researchers to determine causal effects and magnitudes associated with independent variables as they relate to dependent variables at specific points in time. In most cargo data at JAXPORT, an ordered series was observable. Still, in others, it was not.

In lay terms, we presume data are reflective of a true series; that is, there exists a cyclical component within the data of some form. However, that cycle need not be linked to yearly or monthly time intervals; cycles may occur every seven months, or every 13 months, for example.

For commodity data at JAXPORT, cycles varied across each series, but generally we considered the series to occur in 12 month cycles. Oddly, however, a few series exhibited five and six month cycles. In any case, multivariate time series analyses yielded robust models, most of which exhibited strong difference factor relationships; and model–and model components, i.e. intercepts, slopes–that were statistically significant at established alpha levels. Of course, such did not hold true across every series. Nonetheless, we concluded that when applied correctly, multivariate time series analysis has the potential to result in robust projections that can provide sound estimates of inbound and outbound cargo across commodities.

About the Author

Herbert M Barber, Jr, PhD, PhD is a respected author, engineer, economist, researcher, and expert in financial and economic performance of large infrastructure investments. Over the last 30 years, Dr. Barber has provided advisory and consulting in engineering economic systems as it relates to the implementation of large economic endeavors in industry and infrastructure across multiple countries. He is a seasoned scientific researcher with a keen understanding regarding the statistical and econometric effect and causality large financial and economic endeavors have on companies, governments, industries, and economies in developing and developed economies.

About Xicon Economics

Xicon Economics brings intellectual rigor, objectivity, and real-world experience together to solve complex engineering, economic, and financial problems in an effort of increasing financial and economic output. Whether calculating the economic and financial feasibility of constructing a new high-speed rail system, analyzing policy changes in various regulatory agencies, mining smart grid data to develop real options valuations, or developing advanced energy algorithms, we stand ready to leverage our backgrounds economics, research, and statistics to solve complex problems.

 

Share This:


Share This: