Hypothesis 2B. If entry is driven by the entry of uninformed investors, then the number of entry will be positively related to bid premium
4. Data and Methodology
My sample includes five-hundred-ninety-seven public tendering of state-owned land in Taipei City and Taipei County during 09/2007–02/2010. I obtain detailed bidding information on each auction from the website of National Property Administration of the Ministry of Finance, number of bids and the bid price of every bid. The dataset also includes information on the land size, the reservation price, land use code, and whether a building is attached or not. Of the five-hundred-ninety-seven land auctions in my sample, two-hundred-ninety-five were successful.
Table 1 presents the descriptive statistics for the land auction sample. Panel A displays the statistics for the entire sample. The average tendering has a land area of 373.23 thousand square meters, a land area plus building area of 431.24 thousand square meters. For a typical successful auction, the average number of bids is 6.91 with a standard deviation of 7.95, and the average bid premium (the ratio of winning price over the reservation price) is 1.50 with a standard deviation of 1.84. Consistent with Sherman’s (2005) model prediction, there are substantial variations in the number of bids and in bid premium.
Panel B of Table 1 displays land auction background information by year. The average number of bid increases from 2007 through 2010, indicating that land auction has become more and more popular as we see market return has also been increased.
Another interesting factor, the average bid premium, seems to have a decreasing
trend.
Table 2 presents summary statistics of bidding activities in Taipei City and in Taipei County. We can see the successful ratio in Taipei City (205/393) is higher than that in Taipei County (90/204), and the average number of bids in Taipei City (7.12) is significantly higher than that in Taipei County (6.41). We also see the average land area in Taipei City is smaller than that in Taipei County. Market return calculated based on the housing price index shows that Taipei City had a higher growth rate during the sample period. The competition in Taipei City seems more intensive than that in Taipei County. I predict that bidders in Taipei City’s land auctions will be more likely to expend resources producing a more accurate valuation of the land.
I use bid premium to measure the aggressiveness of bids or the level of optimism of bidders regarding the land value. The bid premium is the ratio of the winning price over the reservation price. The Wilcoxon signed rank test for differences in medians between Taipei City and Taipei County shows that bidders in Taipei City pay higher premium. I have more discussion on this in the bid premium regression analysis.
I will examine the factors that influence bidders’ entry decisions. Hypothesis 1 predicts that entry will be higher when information uncertainty or information costs are lower. I have two tests on Hypothesis 1. The first one is a logistic regression to test what factor affecting the success of a land auction:
i i i
i i
i i
i
PPM LA BD RD RM CD
S
=β
0 +β
1 +β
2 +β
3 +β
4 +β
5 +β
6 +ε
(1) The dependent variable, S, is a dummy variable and equals 1 if the auction is successful and 0 otherwise. I use a set of variables to measure information uncertaintyI therefore include only a dummy variable, BD, to indicate whether there is a building attached to the land. RD is another dummy variable equaling 1 if the land use code is for residential use and 0 otherwise. I include two variables to measure investors’
expectation and market condition: one is the previous auction’s bid premium and the other is the housing index return. To explore how bidders are affected by previous auction result, I include previous auction premium, PPM, into the regression model.
Market returns, RM, are calculated based on Housing Price Indexes in Taipei City and in Taipei County. I also include a dummy to distinguish the location in Taipei City and in Taipei County. I run this logistic regression for all sample, and for sample in Taipei City and in Taipei County respectively.
I also test an entry regression using the natural logarithm of the number of bids as the dependent variable:
i i i
i i
i i
i
PPM LA BD RD RM CD
N
) =β
0 +β
1 +β
2 +β
3 +β
4 +β
5 +β
6 +ε
ln( (2)
According to Sherman’s (2005) information production model, number of bidders will decrease with higher information uncertainty or information cost. I conjecture that the information costs and/or uncertainty about the land decreases with larger land size, building attached to the land, the residential use code, higher previous auction premium, and higher market returns. I run this entry regression for all sample, and for sample in Taipei City and in Taipei County respectively.
I then examine bidder aggressiveness on land auctions. Especially, I like to know how bid premium is affected by the entry of bids. Since entry and bid premium are both exogenously determined by auction characteristics, I apply structural equation model to estimate how entry can be affectd by different auction characteristics and how bid premium is affected by entry and other auction characteristics. Structural equation modeling (SEM) is a methodology for representing,
estimating, and testing a network of relationships between variables (measured variables and latent constructs).
The relation among bid premium, number of entry, and auction characteristics can be describe by Figure 1 as follows:
Figure 1 Path analysis of bid premium, entry and auction characteristics
The structural equation model is as follows:
i The dependent variable, BP, the bid premium, is the ratio of winning price over the reservation price. .
Hypothesis 2 says that if there are more entry of informed bidders, bid premium will be lower. On the contrast, when there are more uninformed bidders enter the auction, bid premium will be higher. To check which result holds after controlling for other characteristics that are known to influence bid premium, I apply structural
and in Taipei County respectively.