Suggestions for statistical method and good literature

#1
Hi all,

Im currently doing an assignment where im implementing legal proposals that affects the valuations on commercial real estate on the Danish market. Im utilizing DCF valuations when doing so, but it would be nice to also do some statistics on top of the valuations to see which aspect within the legal proposals has the greatest impact. And because my Supervisor thinks it could be great with some more theory.

The problem is, that its many years since i had statistics and only on a very generel BSc level, where we only did Multiple linear regressions and im not sure this is the correct method. Ive discussed the matter with my supervisor, but his also a little novice within statistics as its not so important in his field of research.

At present i was thinking to do OLS as i've encountered this method in many researh papers and my Supervisor also thinks its the way to go.

I will shortly summarize my problem as i've envisioned it so far:

Dependent variable:

y = Value

Independent variables:

x1 = Whether a 5 year quarantine period has been applied: Dichotomous variable (Yes/no)

x2 = Cancelation of the quarantine periode through energi renovation: Categorical (No / Increase energi certificate by 3 levels / Spend 3.000 DKK on energi renovations)

x3 = Rent cant deviate by a margin of 10% from the legal level: Dichotomous variable (Yes/no)

x4 = Current energy certificate: Ordinal where one is omitted (A,B,C,D,E,F,G or 1,2,3,4,5,6,7)

x5 = Certain paragraph units cannot be modernized: Dichotomous (yes/no) - there are 4 different paragraph units: Cost driven, partly modernized, fully modernized and unregulated. This applies to the partly modernized that most likely cannot be fully modernized any more. And the goal for a owner is to fully modernize all units to maximize rent.

The goal of the legal proposal is to limit the scope of increasing rent levels through modernizing individual units. Therefore, whenever property changes ownership a quarantine period is encountered of 5 years before the new owner can modernize units and only if the energy certificate is at least C. This quarantine period can be cancelled by either increasing the energy certificate by 3 levels or my using 3000 DKK per gross sq. m. residential area.

Furthermore, that the rent cannot deviate my 10% applies to all methods.

I have therefore done DCF's on the properties as the rules are now and implemented the propsals for the properties by either quarantine periode or energy improvement to see which decreases the value the least. Furtheremore, i have have expensed and increased the energy certificate to C where applicable just to be able to modernize units.

I will therefore end up with around 7 valuations for each property: 1 orginal, 1 with 5 year quarantine, 1 with cancellation of quarantine, 2 with cancellation and quarantine and that rent cannot deviate on both, 2 with cancellation and quarantine where partly modernized units cannot be fully modernized and that the rent cannot deviate with 10%. The reason why the 2 last one is applied for independent (partly modernized cannot be fully modernized) is because there is high unceartainty about what is interpretated as partly modernized, so i might also omit these from the analysis.

I will furthermore analysize 30+ properties to see the effect on a diversifed portfolio of properties. Im also doing ordinary sensitivity analysis: Does the WACC have an effect on the effect of the proposal, what effect does churn have (properly alot since units can only be modernized when they are vacant) etc.

So basically any suggestions on an applicable statistical method would be highly appreciated and also suggestions for great litterature. I prefer litterature that also explain how to use a given software. When i had statistics we used Stata, but i have no preference as long as the litterature is great. My current state book looks a little basic though.

Really hope that i can be pointed in the righ direction!! And thanks in advance for any suggestions :)
 

noetsi

Fortran must die
#2
No one here probably knows what DCF is so you might explain that. :p

What is your dependent variable. How is it measured, nominal, ordered....). Its impossible to answer your question without knowing this. Statistics is not simple, its a bit dangerous to do it when you have little background in it. For example, are you going to test the assumptions of your method (and how). Not doing that is very dangerous, but it is not simple to do. You need to look at something basic, like say the Sage monographs on regression.
 
#3
DCF = Discounted cashflow analysis: This can be used to estimate the value of say a company or in this use the property value/price by estimating the cash flow in the a budget period - here 20 years and then discount this with the weighted cost of capital (WACC). The basic assumption here is that money now is worth more than money in the future hence the discounting with the WACC. A higher WACC means higher discounting/punishment of the cashflow. By utilizing this method for property valuation it allows one to take into account various changes in the cashflow and thereby see the impact on the property valuation e.g. how the valuation will react to having a quarantine period before it can be increased or how the valuation instead would change if we pay (increases energy level) to get rid of the quarantine period.

Thanks for pointing out that i should properly explain this ;)

My dependent variable is the expected price of a given property, so as i interpret it, it is continious and cardinal.

I was also thinking that it could be a bit dangerous, because my main experience is from reading scientific articles where they utilize stats. But on the other hand i have some time to invest in the matter, but would be nice to be pointed in the right direction also in terms of litterature thas coupled with certain statistical package e.g. Stata or SPSS. I would properly also want to test the assumption as you indicate :) I also now that i need to test for multicollinearity e.g. by using spearmann's matrix.

Additionally i also know for sure that the various proposals has an effect as i can deduce the effect from my valuations i have done already, but was thinking to maybe add some more independent variables e.g. proportion (%) of business sq. m. that i see indication on could have an effect and maybe also the propertion (%) of units thats already modernized. Just writing the initial thread got me thinking more.

And if cant seem to make ends meet with the statistics i could depart from this and rely solely on sensitivity analysis, but could be nice with some more theory in the assingment :) At least i would have learned some more :)