< Back to all articles
Why Our Model is Conservative
February 11, 2022
People often joke, “I have yet to meet anyone whose underwriting is aggressive”. While it is true that conservative underwriting is a hackneyed claim and almost paradoxical at this point in the cycle since underwriting or pro formas by nature must be aggressive in order to entice investors into a deal. If a pessimist were to truly underwrite conservatively, how could he or she ever end up being the highest bidder for an asset he or she clearly isn’t optimistic about? Anyway, I’m glad to tell you that my goal today is not to prove to you that my underwriting is conservative (although it is ;)) but rather that my underwriting model itselfis conservative. What I mean by this is, the logical construction and formulas driving the calculations of our discounted cash flow models are conservative.
In no particular order, here are the myriad reasons why the construction of our underwriting model is conservative.
Almost all business plans are predicated on the ability to increase revenue and reduce expenses. However, most models have a less-than-robust method to model out the stabilization period to get from in-place performance to pro forma.This can make it very difficult to buy assumptions such as reducing vacancy from 14% to 5% in month 1.
In our model, we use a few different inputs in order to capture the stabilization/reposition picture more accurately. The inputs are stabilization timeline (in number of months), going in vacancy rate (the initial vacancy rate upon takeover), and of course, the stabilized assumptions. All of these inputs/assumptions are intertwined. For example, for the vacancy rate calculation, the model obviously uses the going in vacancy rate as the month 1 rate. Next, the model uses the following calculation: ((going in vacancy rate – stabilized vacancy rate) / number of months to stabilization) to determine the incremental decrease in vacancy for every month. This allows us to project the real life lease up process of a reposition play. For heavier value-add business plans there is typically a drop in occupancy at takeover, then a steady rise throughout the value-add process. This same math is applied to Other Income, Concessions, Bad Debt, and Loss to Lease. For these inputs, the T12 figure is used as the starting point and the incremental changes are based on the stabilization rate.
Once the property is stabilized, only then is the annual rent growth assumption applied to the rents (this means that we are projecting 0% market rent growth throughout the stabilization period and are relying 100% on the value-add improvements to increase revenue). Furthermore, Other Income (this is typically made up of RUBS, application fees, late fees, pet fees, parking fees, etc.) is grown at the expense growth rate which is typically 2% rather than the rent growth rate which is typically 3%. This is because the line items which make up Other Income is typically not able to be increased as aggressively as rents and are more tied to inflation. Lastly, property taxes have a separate annual growth rate assumption which is typically 4 to 5%, especially in Texas. This is often overlooked but is extremely important since property taxes is usually the largest operating expense in Texas and increases significantly through a value-add business plan since the assessor’s office knows you are increasing the revenue of the property and are thus able to pay more in taxes. To more conservatively account for this, we also use the reassessed property tax value immediately in month 1 (this can be changed for more complex or county specific tax situations).
Our model includes replacement reserves above the line (not as capex) thereby reducing the projected NOI throughout the investment. Not only are these replacement reserves accounted for as operating expenses, they are also assumed to be fully spent. Depending on the deal and debt structure, this is very unlikely to occur because value-add deals have a lot of upfront capex work budgeted which reduces the likelihood of large future capital expenditures. Furthermore, it is largely our intention to sell or at least refinance once the value-add business plan is fully completed, thus limiting the risk of unanticipated capex eating into replacement reserves.
Another nuance of including replacement reserves above the line is the way it affects terminal valuation, or sale value. For example, if you have $300/unit in replacement reserves for a 200-unit property, then you are effectively reducing NOI by $60,000. If this property is valued at a 6% exit cap, the replacement reserves being included above the line reduce the projected sale value by $1,000,000! If the original projected sale price is $90,000/unit before considering reserves, then the $1,000,000 swing results in over a 5.5% reduction in value. Many people could overlook this, thinking it is a minor detail of the overall construction of an underwriting model. However, conservatively accounting for replacement reserves not only reduces projected cash flows but materially impacts valuation.
Similar to the above discussion of conservative terminal valuation, there are many ways to calculate your projected sale price. In a relative valuation model, the exit price is calculated by dividing NOI by a terminal cap rate. The terminal cap rate is often one of the first things asked by even the most beginner investor so there is no hiding there. However, how you calculate the NOI used in the terminal valuation is where things can get interesting. Again, as you saw in the replacement reserves example, a small detail of omitting replacement reserves from your operating expenses could result in overstating your projected sale price by over 5.5%, or $1,000,000. Additionally, most models use the year of sale NOI as the terminal NOI, however, some models aggressively grow the year of sale NOI by an additional year’s worth of rent growth and use that as the terminal NOI!For example, if you were to project a sale in year 5, you could use year 6 NOI instead, which could be 2 to 3% higher than year 5 depending on the assumptions of the model. While 2 to 3% higher NOI doesn’t sound all that nefarious, the resultant 3% increase in projected sale price affects the net of fees IRR by approximately 7%, which would be an overstatement of returns. This is the returns for the equity piece of the project are more sensitive to exit value than the project itself would be on an unlevered basis. Some may argue that future buyers of the asset will underwrite their year 1 NOI and factor that into their offer price calculation, thus supporting the year of sale plus one theory. However, this is countered by the fact that future buyers are also anticipating a higher operating cap rate than their purchase cap rate and therefore if you are a believer in the “exit year plus one” theory, you would also then have to increase your terminal cap rate to match the buyer’s year 1 pro forma cap rate. As you can see this line of reasoning becomes tenuous and potentially convoluted which is why it is much more logical and honest to underwrite the sale price on a trailing, in-place NOI basis.
For bridge loans with an interest reserve, our model assumes the interest reserve gets fully utilized even if the cash flows can support the debt service on its own. The interest reserve being fully utilized rarely occurs and the balance is typically returned to investors upon sale or refinance. Next, while we structure our bridge loans so that interest is only charged on funded capex, our model assumes interest is charged on the full bridge loan amount starting in month 1. It could be argued that this is not conservative but in fact just lazy or inaccurate and I’d probably agree with you. Due to the thinking and analysis that went into writing this, I’ll now consider changing my formulas to incrementally increase the interest payments on the bridge loan as capex is drawn (which can be easily modeled on a linear timeline determined by the stabilization timeline input). Lastly, (and this one is less about constructing conservative models and more about showing off) there is a LIBOR forward curve in the pro forma which projects out LIBOR over the next 10 years, giving our model the power to project out floating rates based on today’s LIBOR forward curve.
Franchise taxes should be included above the line. Sometimes this is completely missing or below the line (I’m not sure how it could ever be considered below the line).
Property tax and insurance escrows made at acquisition with the lender are considered a “cost” in the model and are not refunded back to the deal upon sale. This again could be considered more of an inaccuracy than conservatism but I like it for its simplicity.
Operating capital is reserved upfront in our acquisition uses which is typically equal to two months of operating expenses. This is not entirely related to model construction and more a function of the general business plan.
At the end of the day, you should always thoroughly underwrite an investment yourself using your own model (or one you understand and trust) rather than rely on any underwriting that is provided to you. Furthermore, it should be less about the exact percentage return, such as requiring a 15% minimum IRR, but rather the projected return should fall into the top 5% of results for all of the deals you come across. Doing so will ensure the investment’s merit is strong irrespective of how the numbers were calculated, as long as they are calculated consistently.