While it is generally recognized that real GDP, real personal income and total employment are the major economic drivers for the growth in room night demand at the Metropolitan Statistical Area (MSA) level, little is known about the type of metro area that has a major impact on the growth of hotel room demand and ultimately RevPAR. We have therefore used multiple correspondence analysis (MCA), a technique for displaying multivariate categorical data graphically by deriving coordinates to represent the categories of the variables involved, which may then be plotted to provide a “picture” of the data.

Let’s examine the relationship between the Top 25 hotel markets in the US and the metro classifications designated to each of the markets by Moody’s Analytics. The Economic Drivers – Metro Classifications used by Moody’s include: Logistics, State Capital, Manufacturing, College Towns, Defense, Energy & Resources, Federal Government – Nondefense, Financial Center, Medical Center, Retiree Haven, Tourism Destination, High Tech and Agriculture.

Each of the metro classifications are defined by Moody’s. For example a Medical Center is defined as follows:

“Metro areas in this category have a relatively high share of employment at general and specialty hospitals, as well as medical laboratories. Their long-term growth prospects are excellent, largely because of the aging of the U.S. population. As an increasing percentage of the baby boomers become senior citizens, their demand for healthcare will steadily increase, with these metro areas being the primary beneficiaries. Medical centers are identified by the hospital and medical labs employment location quotient. Hospital and medical labs employment is defined as the aggregate of employment in the following industries, as defined under the North American Industrial Classification System:

6221 General Medical and Surgical Hospitals

6222 Psychiatric and Substance Hospitals

6223 Specialty (except Psychiatric and Substance Abuse) Hospitals

6215 Medical and Diagnostic Laboratories”

Each MSA is assigned two to three metro classifications depending on the type of employment and the industries with sizeable concentrations in the MSAs. For example Anaheim is assigned to three classifications, Medical Center, Tourism Destination and High Tech. Los Angeles is assigned, Logistics, Medical Center and Tourism Destination and New York is assigned Financial Center, Medical Center and Tourism Destination. The number of cities by metro classification is shown below.

State Capital2
College Towns2
Energy & Resources1
Federal Government-Nondefense1
Financial Center13
Medical Center8
Retiree Haven2
Tourism Destination11
High Tech11

With twelve metro classifications and twenty-five markets it is difficult to visualize how the MSAs (rows) and metro classifications (columns) are related. MCA allows us to map the rows and columns of a table into a reduced space so that each row and each column will be a point in this joint map. MCA is essentially a dimension-reduction tool that produces a map where each row is at the centroid of the columns it is associated with and each column is at the centroid of the rows it is associated with.

The rows that are placed close in the map have similar column profiles while the columns that are placed close to each other are associated with similar sets of rows. The derived correspondence analysis coordinates are used to plot the MSAs and the metro classifications as illustrated below.

Each MSA is mapped in the same space as the twelve metro classifications. Two MSAs that are close to each other have similar metro classifications. The following graph plots the markets and the metro classifications with dimension 1 and 2. The two dimensions represent 36.5% of the variation in the chi-square table and three dimensions represent over 50% of the variation. The graph illustrates that San Diego, Norfolk-Virginia Beach and Oahu share some common features when it comes to Defense while Phoenix and Tampa share some common features when it comes to Retiree Havens.

Correspondence Analysis Solution for Location of the Top 25 Hotels Markets along Federal Government/Nondefense – Retiree Haven (Dimension 1) and Federal Government/Nondefense – Energy & Resources (Dimension 2)

Source: Hotel Investment Strategies, LLC based on Moody’s Metro Classifications

Correspondence analysis is a useful technique for investigating relationships between categorical variables. By displaying these relationships graphically, MCA allows us more insights than would be obtained from, say, a single chi-squared statistic testing for independence. The analysis can be used to cluster hotel markets that may have similar economic drivers. Such information can be used for diversification purposes and assigning personnel with insights into the industries associated with the major metro classifications.

We have used the technique to assist clients in targeting hotel markets for investment purposes, examining guest perceptions on competing resort brands, identifying latent dimensions of service usage and to map customers and services on a common latent space.

Correspondence Analysis Solution for Location of the Top 25 Hotels Markets along Federal Government/Nondefense – Retiree Haven (Dimension 1) and Retiree Haven-College Town (Dimension 3)

Source: Hotel Investment Strategies, LLC based on Moody’s Metro Classifications

Mathematically, MCA can be regarded as either:

  • a method for decomposing the chi-squared statistic for a contingency table into components corresponding to difference dimensions of the heterogeneity between its rows and columns. or
  • a method for simultaneously assigning a scale to rows and a separate scale to columns so as to maximize the correlation between the resulting pair of variables.