i am working on historical return on direct real estate and I need to make a forecast base on the established return data to have a precise prediction, can I be guided by you in this regard?
Your question is quite broad. By direct real estate returns, I will assume you mean some type of commercial real estate asset? Predictive regressions in RE are generally of a log hedonic form, ln(sales price or sales per square meter or foot)= intercept + building attributes+ locational controls + time controls + other.
Depending on the time series and data you may need availability of capital and/or interest rate controls. You may need green building controls.
I also developed a hedonic grid method that uses a traditional appraisal technique augmented by localized parameter estimates from hedonic regression. That paper is in the forthcoming section of Journal of Real Estate Research.
I suggest you look for articles on your property type in JRER, JREFE and REE, you might also look at Journal of Real Estate Portfolio Management as a lot of cycle and predictive lit ends up there.
This paper discusses real estate equity returns but more recent studies may be more appropriate. You can find other studies by using Google Scholar and find papaers that have cited AKM.
Asabere, Paul, Robert T. Kleiman, and Carl B. McGowan, Jr. "The Risk-Return Attributes of International Real Estate Equities," Journal of Real Estate Research, Summer 1991, pp. 143-151. [Abstracted in CFA Digest and ISFA Digest]