5. Analysis of the Results
5.1. Bivariate Relationships of CEPs
5.1.2. Macroeconomic Variables
國
立 政 治 大 學
‧
N a
tio na
l C h engchi U ni ve rs it y
Figure 5-8 CEPs Year Difference and R&D Employees in California One-Year Difference
5.1.2. Macroeconomic Variables
Overall, most of the macroeconomic variables used in the regression matrix were statistically significant and also not found to be spurious when other variables were introduced.
The impact of macroeconomic variables on CEPs differ in their strength, but for the most part it appears to be true that macroeconomic variables can overall considered to impact the price of CEPs significantly. Again, these include a proxy for supply (industrial production of business equipment), a proxy for demand (the ratio of inventory over sales), GDP, NRCS, and the monetary policy indicator of the effective federal funds rate.
‧
國立 政 治 大 學
‧
N a
tio na
l C h engchi U ni ve rs it y
The smaller number of observations in regressions which used data processing CPI also impacted the strength of the relationship between several macroeconomic variables and construction equipment CPI. This was most obviously the case with the ratio of inventory over sales of equipment where the coefficient increased from around 4 to around 23 for nearly every set of equations holding constant and not holding constant data processing CPI. This can be interpreted as the relationship between the ratio of inventory over sales and CEPs is much stronger during the past two decades as opposed to the last three decades. As can be seen by the figure below in Figure 5-9, the coefficient between construction equipment and the ratio of inventory over sales from 1993 until 2000 would likely be negative whereas the coefficient including data from 1993 until 2019 and data from 2001 until 2019 are both positive with the latter coefficient enjoying a much higher figure.
‧
國立 政 治 大 學
‧
N a
tio na
l C h engchi U ni ve rs it y
Figure 5-9 CEPs One-Year Difference and the Ratio of Inventory Over Sales in Manufacturing One-Year Difference
The reason for why this is the case is somewhat outside the bounds of this research, but it may be deduced that this change in the directionality of this relationship could be related to either an increase in technological development, an increase in speculative appetite for the construction industry, or an increase in either supply or demand for construction equipment. Another option is that there is really no substantive relationship between these variables at all and it just happens to be the case that they are correlated with one another, but this is less likely given the fact that the ratio of inventory over sales is a strong proxy for demand. We know this is the case because of the strong linear relationship between this specific proxy for demand and the model’s proxy for supply which is the industrial production index for business equipment. This linear relationship between
‧
國立 政 治 大 學
‧
N a
tio na
l C h engchi U ni ve rs it y
the model’s supply and demand proxies is demonstrated in Figure 5-10 using data from 2001 until 2019.
While it is helpful to conceptualize this variable as true demand, in practicality it is more accurate to assert that this variable is supply relative to demand. However, it is the supply of only manufactures versus the demand of either wholesaler markup companies or direct consumers (usually contractors) of machinery and equipment. Either way, the positive relationship between these two variables is in concordance with the theoretical models that were presented in the literature review. Moreover, this variable, since it is relative manufacturing supply over demand, is more robust than what a variable simply measuring sales would suggest since it provides a relative measurement of the consumption of capital goods as opposed to a mere nominal figure.
‧
國立 政 治 大 學
‧
N a
tio na
l C h engchi U ni ve rs it y
Figure 5-10 Industrial Production of Business Equipment One-Year Difference and the Ratio of Inventory Over Sales One-Year Difference
As demonstrated in Table 4-1, the results show strong evidence for a statistically significant relationship between CEPs with the model’s proxy for demand and also since there is a strong relationship between the model’s proxy for demand and the model’s proxy for supply (see Figure 5-10). Because of this, it is not surprising then that there is a consistent statistically significant relationship between the proxy for supply (industrial production for business equipment) and CEPs. However, unlike the relationship between the proxy for demand and the proxy for supply, the relationship between industrial production of business equipment and CEPs is much weaker than the relationship between the ratio of inventory over sales and CEPs. The regression results consistently show that the coefficient between the two variables in each model is quite small.
‧
國立 政 治 大 學
‧
N a
tio na
l C h engchi U ni ve rs it y
Moreover, in one of the models where technology and NRCS are held constant, the industrial production of business equipment is not even statistically significant with the dependent variable.
Therefore, it seems to be the case that supply plays a much smaller role than demand in the determinants of construction equipment price changes in the long-run. While this runs contrary to the theoretical literature, it appears to be the case that the US construction industry is an oligopoly where only a few producers control most of the market (Milford, 2015)
Another macroeconomic indicator variable that was consistently statistically significant and appeared to have a strong impact on CEPs was the monetary policy variable of the effective federal funds rate (see Figure 5-11). Similarly to the relationship between CEPs and the proxies for supply and demand, it appears to be the case that a recession can differentiate the relationship between these variables. The 2001 Tech Bubble appeared to be a strong disembarkation point between different correlation strengths with the supply and demand proxy. It appears to be the case that this point in time was a strong divider between a weak correlation of the effective federal funds rate and CEPs as well. In other words, the time period before the recession saw a weaker relationship between these variables than what it did afterwards.
‧
國立 政 治 大 學
‧
N a
tio na
l C h engchi U ni ve rs it y
Figure 5-11 CEPs One-Year Difference and the Effective Federal Fund Rate One-Year Differenced
This then begs the question of why this is the case. What is it about monetary policy in the pre-2001 Tech Bubble that differentiated it from the post-2001 Tech Bubble? A likely supposition is that the Tech Bubble increased efficiency in the tech sector, consolidated the marketplace, weeded out the bad companies with dramatic over-valuation, and increased technological throughput or the successful rate of which digital information is transmitted and processed (Pettit, 2007). Because of this, the Tech Bubble was arguably quite healthy for economy since it reinserted correlations that break down during times of irrational exuberance. Because of the increased equilibrium accuracy of pricing in financial markets and because of the fact that prices for many marketplaces are cointegrated, it could be the case that financial market pricing could be one of the more important factors when it comes to capital market prices. We can also make this