Italian lease events changes in tenants behaviour in

Italian lease events: changes in tenants’ behaviour in the first year of the crisis Lease expiries, breaks and void periods in 2008 Luigi Pischedda - Country Manager Italy, IPD Duccio Martelli - University of Rome “Tor Vergata” Claudio Giannotti - University LUM Casamassima (Bari) On the pulse of the property world

Agenda On the pulse of the property world Ø Italian lease events 2008 Ø Databank features Ø Executive summary Ø Expiry analysis Ø Break Analysis Ø Void periods Ø Model Ø Introduction Ø Hypothesis Ø The regression function Ø Preliminary results Ø Strengths and limits of the model Ø Next steps www. ipd. com

Databank features On the pulse of the property world The IPD Italian Tenancy Universe 2008 l As at the end of 2008, the databank included circa 8, 000 leases, for a total rental value of over Euro 2 bn. l The total market value of the 1, 762 properties was circa Euro 21 bn. Sector weights by number Sources: IPD Sector weights by rental value www. ipd. com

Executive summary On the pulse of the property world 1. Lease expiries What was the result of leases expiring in 2008? Sources: IPD Weighted by number Weighted by rental value Renewals (%) 52 56 Re-let (%) 27 4 Vacant at year end (%) 22 40 www. ipd. com

Executive summary On the pulse of the property world 2. Break clauses What was the result of leases containing a break in 2008? Sources: IPD Weighted by number Weighted by rental value Break not exercised (%) 50 39 Exercised and re-let (%) 43 46 Exercised and vacant at year end (%) 7 15 www. ipd. com

Lease expiry analysis On the pulse of the property world Sectors, % weighted by number Sources: IPD www. ipd. com

Lease expiry analysis On the pulse of the property world Sectors, % weighted by rental value Sources: IPD www. ipd. com

Break analysis On the pulse of the property world Sectors, % weighted by number Sources: IPD www. ipd. com

Break analysis On the pulse of the property world Sectors, % weighted by rental value Sources: IPD www. ipd. com

Void period analysis On the pulse of the property world Segments, % weighted by number, 2008 Sources: IPD www. ipd. com

Void period analysis On the pulse of the property world Segments, % weighted by number, 2007 Sources: IPD www. ipd. com

Preliminary considerations On the pulse of the property world • The global economic and financial crisis which hit the Italian real estate market in 2008 is reflected in the lease events occurred in the same year. • The comparison of 2008 figures against 2007 results show a deteriorated lease market in terms of lease behaviour following the expiry of the contract or a break. • The void period figures of vacant units in 2008 has deteriorated: while in 2007 most of the vacant units at year end (87% of the total number) was in the 0 to 6 months interval, in 2008 nearly 60% of vacant units had a void period within the range 6 -12 months. www. ipd. com

Agenda On the pulse of the property world Ø Italian lease events 2008 Ø Databank features Ø Executive summary Ø Expiry analysis Ø Break Analysis Ø Void periods Ø Model Ø Introduction Ø Hypothesis Ø The regression function Ø Preliminary results Ø Strengths and limits of the model Ø Next steps www. ipd. com

The forecasting model On the pulse of the property world Introduction • Tenant risk has reached higher importance especially in recent years, due to markets falls and consequently the increase in the number of break option exercises • The aim of this model is to investigate the critical factors affecting the probability of leaving a building vacant by the tenant before the natural expiration date of the contract in the Italian real estate market • The model which is going to be presented is a preliminary version: all comments and suggestions are very appreciated • Nowadays only very few models try to estimate the probability of leaving a building vacant; probably this is mainly due to the absence of available databases • The study is based on IPD Italian databank and thus the factors taken under consideration are mainly economic and contractual www. ipd. com

The forecasting model On the pulse of the property world Hypothesis • We adopted a methodology which is mainly used in medical researches to identify factors affecting diseases: the logistic (or logit) regression • Since tenant risk is a kind of risk like any other, it is possible to apply this model to real estate market • The logistic regression model is a non parametric model where the output variable is binary (assume values 0 or 1) depending if that variable has the characteristic investigated or not • Besides, the logistic model combines the easiness of the construction to the immediateness of results interpretations www. ipd. com

The forecasting model On the pulse of the property world The regression function • The regression has been applied on a sample of about 250 contracts which could be closed prior their natural expiration date during the years 2007 and 2008 • Like for all regressions implying qualitative variables, the model identify a comparable building, which in this case has been defined as an office located in Rome, with a break option dated 2007 • Taking into consideration collinearity problems (an independent variable is mostly explained by another) and estimation problems (absence of the event analyzed in some considered variables), the function used in the model is the following: www. ipd. com

The forecasting model On the pulse of the property world The regression function yesbreak = β 0 + β 1*sector + β 2*sqm 100 + β 3*efferv 100 + β 4*area + + β 5*city + β 6*renov + β 6*year Variable name yesbreak sector sqm 100 efferv 100 area city renov year Description Dummy variable with value 1 if the building has been left by the tenant during the year; 0 if the tenant has not exercised the break option Dummy variable which can assume the following values: Hospitality & Other, Industrial, Office, Residential, Retail The size of the building in square meters (actually sqm/00) Indicates the difference between the effective rent paid by the tenant and the estimated rented value that could reasonably be expected to be obtained on a new letting (on annual basis for 100 sqm) Dummy variable which can assume the following values: Center, North-East, North-West, South Indicates whether the building is located in a big city or not. Dummy variable which assumes value 1 if the building has been recently renovated or 0 otherwise. Indicates the year during which the break option can be exercised. It can assume the following values: 2007, 2008 www. ipd. com

The forecasting model On the pulse of the property world Preliminary results are shown in the following table: How to read the table ü Exponentials of coefficients represent the number of chances (more or less) compared to the comparable building ü The most critical variables seems to be sector of activity, year of exercise and location ü Although the pseudo-R 2 seems to be low, we have to remember that the model considers just economic and contractual variables www. ipd. com

The forecasting model On the pulse of the property world Strengths and limits of the model • Strengths and limits of the model can be summarized in the following points: ü It identifies the key factors affecting the tenant risk, but probably it overestimates the intensity of the relations ü It excludes some variable, not because they are not important, but just because no break options were exercised during the period 2007 -2008 (i. e. industrial sector, specific regions, …) ü It is based on a two-year sample • Although the model is not limits exempted, it can represent a good starting point for future developments www. ipd. com

The forecasting model On the pulse of the property world Next steps • In order to solve these issues, next steps regard the possibility: … ü …to extent the time series used for the analysis ü …to transform the results from “numbers of chances” to “probabilities” of break exercises for every building, no matters if this one is included or not in the sample ü …to extend the analysis to other real estate markets, in order to identify differences in risky factor affecting the chances/probabilities of leaving a building vacant www. ipd. com

The forecasting model On the pulse of the property world Operative implications • In particular, once those issues are solved (or limited), we think that the model can be used especially by landlord for two main applications: I. Identifying risky situations before they really happen (and thus renegotiate the contract before the break option exercise) II. Pricing the contract correctly, linking the rent to the real riskiness of the investment (taking into consideration both exogenous and endogenous variables, but also risky factors related to the tenant) www. ipd. com