Metropolitan Area Markets for Cement,
Concrete & Stone is a multifaceted report. Depending on what you order,
you may receive a report encompassing only a single U.S. metropolitan area or as
much as five thick volumes covering five different building materials --
portland cement, masonry cement, aggregate, ready-mix concrete, and “other”
concrete – for 361 metropolitan areas. This brief guide explains the
methodology used to derive these metropolitan area estimates and forecasts.
High freight costs dictates market proximity
Cement rarely is shipped more than 100 miles
from the plant or distribution terminal to the customer. For concrete and
aggregate, the distances from supplier to customer are even less. Aggregate
quarries and ready-mix concrete batch plants are located near the metropolitan
areas they serve. Portland and masonry cement may be shipped from a plant in an
outlying county, but cement plants typically are within a 100-mile radius of
their main urban markets.
The vast majority of cement is shipped by truck.
U.S. Geological Survey (USGS) data for 2003 indicate that 96% of cement
shipments to consumers were by truck. Rail transport accounted for virtually all
the remainder with a tiny 0.2% shipped by barge. Rail and barge transport are
more important in shipments between cement plants and terminals: 44.1% for rail,
39.6% for barge, and 16.4% for truck.
Shipping cement by truck even 50 miles can add
$10 per ton to the delivered cost, a significant amount for a $80 per ton
product. The freight economics for ready-mix concrete and aggregate are even
steeper as the distance between customer and supplier increases.
What about imports?
Foreign cement imports are an important supply
source for U.S. cement companies, satisfying 25%-30% of domestic demand. Except
for a small percent of cement from Mexico and Canada, imports are shipped here
by ship or barge to domestic cement plants or terminals, frequently traveling
thousands of miles.
Once the cement imports are delivered to the
local cement plant or terminal, they are shipped by truck to the customer, no
different from domestic product. The freight economics over land are the same.
Stick close to home
Cement, concrete, and aggregate all have low
value by weight. Consequently, freight costs rise rapidly as delivery distance
increases. Being close to customers is a significant competitive advantage. Only
premium-priced, specialty concrete products, cements, and stones can
economically travel great distances. These premium products are a tiny
percentage of total sales. The rest must ship to customers near plants,
terminals, and quarries to avoid substantial delivery costs.
Metropolitan area markets are most relevant
Given the short distances that cement, concrete
and stone travel from plant to customer, metropolitan areas are obviously the
most relevant market areas for suppliers. One cement plant may sell throughout
an entire state, but it is rare for an aggregate or concrete operation to sell
beyond the nearest metro area. With access to rail or barge transport, a cement
plant may expand its market greatly using distribution terminals in distant
urban markets, but this is exceptional. For the terminal, moreover, the relevant
market is its immediate metro area.
No metro area data collected
The USGS does a great job of collecting data on
cement and aggregate, but state level information is the best geographic detail
available.
Only national data are collected by the Census Bureau on concrete. Thus,
building materials companies do not have information on their metropolitan area
markets, even though this is most relevant to them. They cannot even compute
their market shares because they don’t know how big the market is.
Furthermore, they don’t know how fast their markets have been growing, or how
cyclical they are. They cannot compare one metro area market to another. All
they can do is assume that what they experience individually as companies is
true for the total market. This is a dangerous assumption. It means you are
assuming market shares and growth rates are constant.
If there are no measures of past or present
metro area market size and growth, how can forecasts be made? Companies need
projections of future growth for budgeting and planning purposes for current
operations. When making acquisitions or expanding into new markets, forecasts
are also critical.
Estimates need to be made
With no actual metro area data collected,
estimates have to be made. What is the best method? I believe Metropolitan
Area Markets for Cement, Concrete & Stone uses the best methodology
available. The estimates in this report were made using a statistical technique
called multiple linear regression (MLR). It makes best use of the information
that is available. MLR models are also known as econometric or ordinary least
squares (OLS) models Goodness-of-fit statistics measure how well the models fit
the data. They are usually referred to has R2, F, or t
statistics.
Econometric models work best when there is a
sound theoretical basis for the functional form of the model. Rather than
randomly picking variables to explain cement (or concrete or aggregate) demand
based on correlations, the explanatory variables should be selected based on a
theory of what determines demand. The theory can be based on esoteric
mathematical models or simply on experience and common sense. The estimated
models should confirm the correctness of the theory by fitting the historical
data well.
The determinants of building materials demand
Building materials demand is generated by
construction activity. A variety of factors influence the level of construction:
interest rates, unemployment, job growth, tax policy, government spending,
regional cost of living differences, etc. Most of these factors influence the
cyclicality of building activity, e.g. booms and busts, expansions and
recessions. Some of the factors are national in influence, e.g. interest rates
and monetary policy. Most of the factors have a local aspect as well as state
and federal, e.g. job growth, unemployment, and taxes. There is competition
among cities, states, and regions for construction contracts. Some areas have a
comparative advantage over others, which benefits construction activity.
All of the above factors are essentially
short-term or intermediate-term influences on the business cycle for
construction. What are the long-term influences on construction? Population is
the most important long-term determinant of construction activity. Population
size and growth determine the need for most types of construction, e.g. housing,
retail, roads, offices, schools, sewers, hospitals, etc. Population is the
primary driving force behind virtually all construction activity over the
long-run.
Technological change is another long-term
influence on construction and other economic activity. Innovation generates new
building needs as well as demand for new equipment and consumer products.
Unfortunately, there are no measures of technological change or innovation at
the state or local level. Even national measurements are “shaky”.
Building state-level models
Some people might build one national model of
cement (or concrete or aggregate) demand and then apply it to every state and
metropolitan area. Decades of experience has shown that national models of
building materials demand perform poorly. There is far too much regional, state,
and local variation in building materials markets to be captured by a national
model.
Econometric models of local markets would be
best, but this is impossible because there are no data collected for cement,
concrete, and stone demand at the metro area level. Thus, there are no
“dependent” variables for the local models, the variables you are trying to
explain. However, relevant “independent” variables (explanatory variables)
do exist for metropolitan areas, namely, population and construction activity.
That leaves state-level regression models as the
best alternative. There are accurate measures of cement and stone shipments from
the USGS, as well as good measures of population and construction activity.
There are no state-level data collected for concrete, however.
Thus, I have estimated regression models by
state for portland cement, masonry cement, and aggregate. That’s roughly 150
equations. While some have better statistical properties than others, all have
good statistical and theoretical characteristics.
The state concrete estimates are derived from
USGS data on the amount of portland cement sold to ready-mix concrete and
“other” concrete customers. USGS reports these data by state. I assume that
a cubic yard of concrete uses 500 lbs. of portland cement, i.e. a ton of
portland cement yields four cubic yards of concrete. Thus, my estimates are
simply four times the USGS data. I realize that the use of fly ash and other
admixtures makes my estimates of concrete use low, but this cannot be helped.
There are no good measures of fly ash use in concrete at the state level.
However, you can adjust my figures based on your own knowledge of fly ash use in
a particular metropolitan area.
Making metro area estimates
The state-level regression models are used to
make metropolitan area estimates of cement, concrete, and stone consumption.
While the process is somewhat more complicated than simply plugging the metro
area population and construction data into its state model, that is basically
what is done.
One complicating factor is model error. The
state models fit the historical data well but not perfectly. Typically, the
models explain 80% to 90% of the yearly variation in state cement and stone
consumption. I spread the error proportionately among the state’s metro areas,
so that everything adds up correctly.
Forecasts
Five-year forecasts of cement, concrete and
stone are also supplied in this report. The basis of the forecast is my annual U.S.
Cement Forecast. This report gives state-by-state portland cement forecasts
for 30 years using three different scenarios: baseline, better, and worse. I use
the baseline forecast in metro area reports. The economic assumptions underlying
the state forecasts are described in detail in U.S. Cement Forecast.
The state-level masonry cement, aggregate and
concrete forecasts are all derived from the portland cement forecasts. The
metropolitan area forecasts are calculated in a similar manner to the historical
estimates with one exception: the metro area forecasts take into account
differences in population growth rates among cities within a state compared to
the state average. Thus, the fastest-growing metro areas in a state have
relatively better building materials forecasts.
Population and construction activity
The explanatory variables for all of these
regression models are population and construction activity. The report gives
20-year historical data for population and residential building. Homebuilding is
separated into single-family and multi-family units. A five-year population
forecast is also presented.
The changing definition of a metropolitan area
The Office of Management and Budget (OMB) is the
U.S. government agency responsible for defining metropolitan areas. These
definitions are revised periodically. New metro areas are added and old ones are
redefined. All metropolitan statistical areas (MSAs) are
defined to include whole counties. Adding a whole county to a metro area can
mean a large jump in population, construction activity, and building materials
consumption.
My estimates carefully take into account all
changes in the definitions of U.S. metropolitan areas since 1979. Changes are
most obvious when a metro area’s population suddenly jumps from one year to
the next. There are significant revisions after each U.S. Census. This happened
to numerous MSAs in 1983, 1993, and 2003. New MSAs are added, some are dropped,
many get additional counties in their definitions, and some shrink.
The calculation of compound growth rates will be
distorted when a MSAs definition is changed. If a county is added, the long-term
growth rate will be artificially boosted. Keep this in mind.
When new metropolitan areas are added by OMB, I
make new estimates of their building materials use. When metro area estimates do
not go back to 1979, it is because the city was not yet designated a MSA by OMB.
Moreover, the Residential Construction unit of the Census Bureau doesn't
immediately incorporate the OMB changes. There tends to be a lag of about a
year.
Summary statistics
The latest edition of Metropolitan Area
Markets for Cement, Concrete & Stone contains 20-year historical estimates and
5-year forecasts. The forecasts for the current year incorporate the most
up-to-date actual USGS monthly state data. For the historical estimates
and forecasts, I have computed annual percent changes as well as two summary statistics:
annual averages and compound annual growth rates. Annual averages give an
indication of the long-run average size of metro area markets. Compound growth
rates are calculated using OLS on the logarithms of the historical estimates.
Conclusion
Metropolitan Area Markets for Cement,
Concrete & Stone is unique. No other product estimates history and gives
forecasts by metropolitan area for portland cement, masonry cement, aggregate,
ready-mix concrete, and “other” concrete.
Please send me your comments and suggestions so
that this report can keep improving.
