# Strategic Risk Management in Real Estate/ a snapshot into Simulations

# Risk Management Support

The idea behind strategic risk management is to **support a decision maker in taking a market position in accordance with the venture’s strategic goals, backed by the current market situation, based on predominating market dynamics, the circumstances of the own portfolio and what can be expected from respective target markets in the future.**

The evaluation of those expectations is therefore data-driven and quantified in terms of — which future scenarios are more probable to materialise.

The practical convenience is the flexibility of implementation (it can be used as SaaS, APIs, Edge solution or even implemented as part of Excel based risk management programs) with 24/7 support.

In order to offer truly advanced analytics as well as quantified approaches and to uphold this support, a whole arsenal of **advanced predictive analytics risk management tools **is** **to be combined.

Here, in this article, we have a look at one of those tools, namely mid- to long-run simulations.

For this purpose, we do not run a whole simulation to get risk/ performance metrics results.

We rather want to see, how much more **flexible and adaptable simulation methods** are (expecially compared to a, e.g., sensitivity analysis in the more traditional business analytics environment) by focusing on one topic in the early stage of a simulation.

In short, we have a look at the **correlation between investment yields in a regional market risk cluster**.

# The Data Set

Starting point of a simulation:

**defining of market parameters**of the asset classes in question,**revealing of their respective market patterns**and- the identification and implementation of related markets in order to reveal the
**impact of a potential cluster risk behaviour.**

All this is ingrained in the interdependencies of the market features represented by the **correlation grid**.

Here is a simplified example of the empirical correlation grid of a regional market cluster consisting of two markets:

Based on this net of associations, we take out one instance, i.e. the correlation among the investment yields of those two markets. The historic development of both yields revealed a **quite strong association**, i.e. so far they showed very similar patterns. The correlation factor equals to 0.8.

So, let’s see how this specific item is taken into account in the overall simulation.

# Investment Yield Correlation

The correlation grid is embedded in the overall simulation which is otherwise fed by the marginal probability distributions of the single market features as well as the joint probability of all the market input as the spanning umbrella.

As mentioned before, the historic development of both yields revealed a quite strong empiric correlation equalling 0.8.

Given this fact (as well as all the other factors), the expected yield development in both markets within, say a 5-years term, is expected to be as follows (the plot below shows some of the simulated scenarios; the dotted black line represents the current market level):

But how does the correlation in those simulated scenarios perform? After all, our empirical baseline was 0.8.

As can be seen in the graph below, the **yield correlation does not statically stay** at the 0.8 — benchmark. **It allows for a much wider range** (see the grey area in the graph) compared to the 0.8 — cut (see the dotted red line).

This seems logical as the **empirical correlation** coefficient of 0.8 is based on the historic yield developments which **could change in the future**. Additionally, that 0.8 is **just a glance into the real market dynamics which we actually do not know**.

**Therefore implementing a wider range of possible correlation factors does reflect a much more realistic approach than sticking to the given benchmark.**

Nevertheless, the **highest probability** stays **with a strong positive correlation** among the investment yields to be realised in future movements. For example, a strong correlation ranging from 0.5 to 0.9 comes with a probability of around 67%. On the other hand, chances to have a correlation factor above 0.9 or even to have a negative correlation are a mere 4% each.

# Adaptable and Flexible

Just keep in mind, **how many potential future scenarios are incorporated** in just this one snapshot of the overall simulation.

And all this potential scenarios are **“weighted” by probabilities**. In other words, at any step you know what are the odds that such a situation might materialise.

Also, consider that we just talked about one tiny piece in the puzzle. In contrast, there are so many more items (as you can see in the first graph/ correlation grid) which follow the same principles.

In such a risk simulation, **all potential scenarios of a market development are evaluated and quantified** in terms of their chances of getting materialised. Those results are **combined with project/ portfolio data in order to derive a data-driven risk/ performance metrics** and to offer “the big picture” **to the decision makers**.

This calculation which is happening in the background is **a very complex statistical modelling approach, but focused on the goal to provide a simple-to-obtain result** — with a single “push of a button” in an application which is available 24/7.

Especially, when you compare the possibilities this method offers with the limitations of sensitivity analysis in a more traditional risk management information system, it gets clear that there is a **new generation of strategic risk models available which changes the “risk mindset” we had in previous decades.**

**Traditional sensitivity analysis is simply limited by the natural boundaries **of creating different scenarios and combining different input features at once. Those traditional analytics miss to link up the different scenarios with their quantification, i.e. answering the question how serious it is that a range of scenarios might occur.

Different scenarios are simply weighted the same way giving the decision maker no real direction, and therefore **not providing the necessary insights to take an informed market/ risk position**.

This small example gives a very good indication **how broad the possibilities and insights are, advanced predictive analytics methods are offering** while **paving the way for a strategic risk management**, which is a complete change of concept we had up to now.

**Instead of hours of manual work with a limited range of results,** now we have a **flexible, practical and fast way** to offer decision makers a strategic tool with an advanced analytical background, using **data driven models in order to detect risks and opportunities — the next generation risk management indeed!**