[Spencer Burton]

Thanks for the questions. Let me do my best to answer them.

1) When does Month 0 start? For the Co-GP entity, does it start when the raw land is taken down? Does this mean that the GP/LP waterfall would have a different Month 0 start? While the gross profit may be the same to the GP entity between the GP/LP waterfall and Co-GP waterfall, before any fees, would the IRRs be different based on a 60-month IRR calc at the GP/LP entity and a 12mth + 60mth period at the Co-GP entity?

Month 0 would generally occur at the GP-LP venture closing. At the property level, any costs to date would be added up and capitalized in time zero (i.e. most often pursuit costs) with the LP typically crediting the GP’s capital account with costs spent to date. If the GP had closed on the land, likewise its capital account would be credited with the value that the GP-LP venture places on the land, which can either be market value or actual cost.

In terms of the second half of the question, you’re right. From the Co-GP perspective, using the the GP-LP venture closing for time zero may throw off time value of money hurdles in the Co-GP waterfall. I have two comments on that.

First, this is one of the reasons Co-GP hurdles usually are simpler and not based on time value of money metrics. You’ll often see a Co-GP structure look something like 8% preferred, non-compounding, with all excess cash flow split 50/50. Then, in your model you’d adjust the ‘Required Return’ for the Co-GP partners in time zero to reflect any accrued preferred return. By the way, many times a Co-GP structure won’t require a preferred return during a period early in the venture in which case even this step would be unnecessary.

[Side bar comment: there’s another more important reason to keep the Co-GP waterfall structure simple, and that’s because the Co-GP investors (“Sponsor + family, friends, and fools’ money”) need you to keep it simple. The saying goes, “people don’t invest in what they don’t understand”. Your aunt May can understand “we’ll split the profit 50/50 at the end”, but she may not understand “90/10 to an 8% IRR, 80/20 to a 12% IRR, and 70/30 thereafter”.]

With that said, if you do have a complex Co-GP waterfall with time value of money hurdles, you would need to run two concurrent models. One to model the LP returns that starts at the GP-LP venture closing, and one to model the Co-GP returns that starts at the Co-GP venture closing. The distributions from the GP-LP model to the Co-GP would then link to the Co-GP model to combine with the contributions from the Co-GP model to accurately model the IRR of the Co-GP partners.

2) From what you’ve seen, how do these deals typically get structure? For example, if the Co-GP entity purchased land for $1.0MM, spent $1.0MM entitling, and believes the value to be $5.0MM, how does the GP/LP entity acquire the land? Does the Co-GP entity sell for $5MM and book a net cash gain of $3MM in that period or do they typically roll the $3MM in equity back into the deal? If they book a cash gain, would you model that at Month 12 (when the Co-GP executes the GP/LP partnership) as a distribution to the Co-GP partners? If they only contribute land, rather than cash, I’m assuming you wouldn’t model any share of construction costs to be contributed by the Co-GP, right?

I partially answered this in question 1, but allow me to elaborate.

Your example is fairly common. The Co-GP purchased the land previous to the GP-LP venture execution and has added significant value to the land through some combination of entitlement/design and/or just simply holding the land for a long time. You’ll often see GPs bring land to a venture that they’ve owned for years (or decades), and thus their cost basis in the land isn’t a fair representation of the value of the land.

In my experience what usually happens is there’s a negotiation over the value of the land. Take your example. The GP has spent $2.0MM to date and asserts that the land is worth $5.0MM. As the LP, I’d need support for that $5.0MM valuation before giving the GP full credit for the land value. But let’s imagine the $5.0MM value can be supported. And further let’s imagine a $50MM total project cost with a 90/10 ownership split. At venture closing, the GP would contribute the land to the venture at a $5.0MM value, and thus would not be required to contribute any further to the venture.

So from a GP-LP modeling perspective, nothing changes. The venture “purchased” the land at $5.0MM and the GP contributed its pro rata share.

From a Co-GP modeling perspective, you’d need to run a concurrent model to appropriately calculate the IRR. And it could get quite complex. But again back to my lecture when I talk about simplifying partnership waterfalls down to cash contributions and distributions. As you model that Co-GP equity waterfall, just focus on what cash contributions were made and when, and what cash distributions were made and when. Ideally you’re building a dynamic model so you can adjust assumptions down the road or use the model on a future deal. But if you get stuck, it’s not the end of the world if the model ends up being static, especially when you’re dealing with once-in-a-career structures.

Happy to answer follow up questions!

[Spencer Burton]

I’m glad you brought this up. As a general practice, I hide any unnecessary rows and columns from my models. I do this for a few reasons.

1) To avoid unintentionally moving my cursor to the far end of the worksheet and make it easier to move around the work space with keystrokes
2) To avoid unintentionally adding data or formatting to the far end of the worksheet
3) I personally like the clean look

So to your question, how do you get access to those hidden rows/columns. Rather than give written instructions, allow me to explain via video.

 

[Spencer Burton]

I hate to answer it depends – but the answer to these particular questions really depends on the borrower, the capital markets at the time, the property type, the in-place capitalization, and a host of other factors. Nevertheless, allow me to offer some general rules of thumb to help provide color for when you’re modeling them.

1) It is best practice to assume refi (or sale) at stabilization, yes. However, you should generally only capitalize (i.e. add to the development budget) interest shortfall, not total interest. So even though you might have 1.5 years of construction interest, much of that interest is paid for by income during lease-up.

2) I wouldn’t. From a modeling perspective, it’s a lot of brain damage for a marginal level of increased precision. A bridge loan to get to stabilization usually occurs when the lease-up pace is slower than initially forecast and the developer needs extra time. But to model that scenario initially seems overkill, especially when the rate and loan amount of the bridge loan is likely to be similar to the construction loan.

2.5) In terms of how to size a loan before stabilization. This is actually quite common. While we generally model refi at stabilization, in the real world the takeout loan is usually underwritten before stabilization and closed when the property is close to stabilized. This is because it takes months to get a loan closed, and so the developer can’t wait until the property is fully stabilized to start the process. All loans, whether the property is stabilized or not, are underwritten to a stabilized pro forma. Then structure , as it’s called, is added to the loan to protect the lender. Structure may take the form of specific conditions to close (e.g. closing once the property has been greater than 92% for 90 consecutive days), or it might take the form of a staged funding (e.g. initial funding with fewer dollars, followed by subsequent fundings as leasing targets are hit. Those subsequent fundings are sometimes called earn-outs or good news money.

3) Whether to underwrite free rent on a stabilized pro forma is always a point of discussion. It really just depends on the property type, market conditions, and lender. For instance, if a soon-to-be-stabilized property with 1st gen concessions is located in a market with zero concessions at other fully stabilized properties in the market, than the lender can make the argument to its IC that concessions at the subject property are likely to burn off. And thus, the free rent won’t likely be underwritten. However, if the property is in a market where concessions are common at fully stabilized properties, the lender will certainly underwrite some long-term concessions into its pro forma.

Great questions!

[Spencer Burton]

Thanks Lv for chiming in on this discussion.

First to answer your question: “Are we not under estimating General Vacancy & Credit Loss.” Depends on who you ask. A lender might argue yes, a borrower would certainly argue no. A buyer might say yes, a seller would certainly say no. This is an imperfect science that ultimately seeks to ascertain (i.e. forecast) a durable long-term valuation, with every assumption at the discretion of the professional making that assumption.

In case of Crescent Apartments (and The Foles at the end of the course), I used a methodology and assumptions that if I were a seller or borrower, I believe I could support and defend. And so if a lender or buyer were to push back, I would argue that I’m actually underestimating revenue (i.e. overestimating vacancy).

Here is what my argument would be.

We’re modeling a forward-looking net operating income. In-place rents at the property are currently 2.6% under market, and effective gross revenue has grown steadily by nearly 3.0% year-over-year since 2016. So if anything our revenue assumptions are conservative, considering we’re only growing EGR by less than 1%.

Now your argument is sound as well. And the lender might make that argument.

In-place physical vacancy is nearly 6%, thus a 5% general vacancy factor is insufficient. But I would point to historical economic vacancy and compare that to pro forma economic vacancy.

Economic vacancy here is calculated as [Loss-to-Lease + Vacancy] ÷ [Market Rent + Total Other Income]. Historical economic vacancy at the property has averaged 4.9%. I’m underwriting 7.15% (143,532 + 309,609 ÷ 5,792,040 + 543,678). Hence the reason EGR barely grows from the T12 to pro forma.

So while a 5% General Vacancy factor on gross revenue might appear to be aggressive, when combined with the loss-to-lease adjustment and compared to the history, it’s actually relatively conservative.

But again to the main point of this exercise. Set 100 real estate professionals to underwrite a direct cap value for a property, and you’ll get 100 different valuations. That’s because every assumption is at the discretion, and influenced by the incentives and objectives of the professional doing the analysis. What’s most important is that you’re thinking about the assumptions, and preparing reasons for why you made each.

The fact you’re digging this deep into these particular assumptions means you’re going to be really good at this!

[Spencer Burton]

Quick addendum.

I should point out that using the ‘Current’ and ‘Accrual’ toggle in real estate financial modeling is not very common. That’s because we generally build models for the property type, market, investment type, and lender convention that we work in. I think it would be unusual to be in a market where your development model needed a toggle to switch between Current and Accrual.

Of course, that’s my perspective and it certainly doesn’t hurt to include the toggle. But more than likely, adding the toggle to the models you build is an unnecessary exercise.

[Spencer Burton]

Hi N2,

Michael teaches you how to model construction interest reserve, so in a way this is covered. But this is a good question to bring up at this point in the Accelerator. I too have seen the ‘Current’ vs ‘Accrual’ toggles in development models. In my experience the terms are used in the context of what source (i.e. debt or equity) covers the construction interest.

In ‘Current’, the borrower pays the interest in the month that the interest is charged. Thus from a modeling perspective, that interest cost is paid for by equity (i.e. the partner’s capital account grows accordingly). In ‘Accrual’, the construction interest is paid for by the debt; or in other words the interest is added to the loan balance each month with interest charged on accrued interest over time – hence the need for an interest reserve.

Now when are you likely to see ‘Current’ interest treatment vs. ‘Accrual’ interest treatment? Keep in mind that, from the lender’s perspective, ‘Current’ interest is less risky because theoretically the borrower has more skin in the game. So whether current or accrual is used depends on property location, property type, profile of the borrower, market norms, etc.

Anecdotally, accrual is the norm in the U.S. on development loans for institutional, income-producing (i.e. for rent) real estate. When I developed single-family in the U.S. that was also the case. However, when I developed single-family projects in Latin America, all banks used the Current interest treatment. Meaning, we had a debt service payment each month during development, rather than that interest accruing.

So at the end of the day, it’s a lender decision driven by market convention and deal risk.

Great questions!

Spencer

[Spencer Burton]

Kyle,

This question inspired me to build a real estate equity waterfall with catch up and clawback provisions included as options. You can check out the model here:

https://www.adventuresincre.com/real-estate-equity-waterfall-model-catch-up-clawback/

At some point, I’ll create a supplementary lecture to this course with a Watch Me Build showing you how to add model the catch up and clawback provisions. I’ll notify you in this thread once that happens.

Spencer

[Spencer Burton]

Thought I’d share an example I quickly put together of how you might model a Catch Up provision.

Click here to download the template used in this video

 

[Spencer Burton]

Another great question!

First an important point on the subject of JV partnership provisions. There are so many variations of these (and other) provisions that it’s impossible to definitively say: “this is how you model a catch up or clawback“. To illustrate what I mean, read Stevens A. Carey’s paper on the subject: ‘Real Estate JV Promote Calculations: Catching Up With Soft Hurdles” where he gives at least half a dozen examples – each that would be modeled differently from the next. And those are just a handful of the dozens of variations you might run into.

So rather than teach you how to model one or two methods, it’s more important to become proficient at modeling waterfall tiers, tracking capital accounts and required returns/distributions, and calculating contributions and distributions accurately. Once you’ve mastered that, you’ll find you’re able to model all sorts of different JV structure provisions without issue.

With that said, allow me to give an overview of the two provisions you bring up and offer high-level suggestions for how to model each.

Clawback Provisions

In my experience, Clawback provisions are most commonly used in JV structures with multiple investments being made by the partnership (i.e. portfolios or multi-phase developments). Clawback provisions are meant to protect the LP from scenarios where later investments underperform earlier investments. So for instance, if cash flow from later investments is insufficient to pay the LP its preferred return and/or return capital, the LP would have the right to “clawback” earlier distributions made to the GP.

Here is a resource you can read on the subject.

Carey also brings up a variation on the clawback concept, even though he refers to it as a “lookback”. He uses it in the context of a single investment, where profits are distributed to the GP before the LP reaches its preferred return hurdle. In the event the investment closes and the LP hasn’t hit its hurdle, the GP would have to return a portion of the profits it has received. I’ve never seen an LP agree to this structure, although I’m sure it happens.

Modeling this involves tracking the cumulative amount distributed to the GP, and using that amount as a cash flow source at reversion from which to distribute back to the LP in the event the cash flow from the property is insufficient to hit the LP’s required return.

Catch Up Provisions

In my experience, Catch Up provisions are most common to pref equity structures or to JV structures where the LP has a priority right to distributions up until it earns its preferred return. Or in other words, in structures where the LP receives 100% of distributions until it hits its preferred return hurdle.

Let’s assume the LP and GP agree that profits should be distributed (i.e. Profits = Total distributions minus amount required to repay contributed capital) 50% to the LP and 50% to the GP until the LP has earned a 10% IRR on its investment. However, the structure requires that 100% of the distributions be made to the LP until the LP has achieved its required return. Thereafter, the GP is entitled to 100% of the distributions until it has received 50% of the profits (i.e until the GP has “caught up” with the LP). Once both the LP and GP have each received 50% of the profits and the LP has earned a 10% IRR, the waterfall proceeds to the next tier.

Modeling this involves tracking the profit distributed and what share is owed to the GP (i.e the GP’s required distribution). Once the LP has hit its required return, excess cash flow is distributed 100% to the GP until it has hit its “required distribution” or in other words until the GP’s profit distributed equals the LP’s profit distributed in that tier.

[Spencer Burton]

First a Couple of Definitions

Let me first say that you’ll see varying definitions of mezz debt and pref equity in real estate. What Michael and I define as “mezz debt”, some people call “pref equity”. So let me make sure we’re speaking the same language.

When I refer to mezzanine debt, I’m referring to “a debt instrument secured by the investment property, but only indirectly, by a pledge of the equity in the entity (usually a limited liability company or limited partnership) that owns the property”. I believe this to be the correct defnition of mezz debt (and the Stanford Law Journal agrees with me). You can also review this journal article on how foreclosure works with a mezzanine loan.

With that said, you’ll often find professionals with a slightly different definition of mezz debt (i.e. here’s one example), so it’s important I define the terms first.

So again, for purposes of this answer mezz debt is a debt instrument secured by the ownership entity. Pref equity on the other hand is a direct equity interest in the ownership entity. However, the pref equity owner is given priority over excess cash flow distributions and usually is only required to contribute its equity share after the common equity owners has made its required contribution.

So How to Model Mezz Debt vs. Pref Equity

So using the above definitions, preferred equity is modeled in your equity waterfall whereas mezzanine debt is modeled in your debt module. Or another way to think about it, mezz debt is a property-level cash flow, whereas pref equity is a partnership-level cash flow.

In terms of how to model preferred equity. It obviously depends on the terms of the pref equity. Typical pref equity in my experience is contributed after common equity, while distributions to the pref equity piece have priority over the common equity.

So modeling the contributions and distributions of the pref equity requires tracking its capital account, contributing from the pref equity only once the common equity has contributed its required share, and then distributing excess cash flow first to the pref equity before distributing to the common equity.

Spencer

[Spencer Burton]

Hi Noam,

Thanks for the question. A variation on this question was asked in our Anatomy of the DCF course. That answer might also be helpful.

https://www.adventuresincre.com/academy/forums/topic/reversion-sales-price-property-tax/

In terms of your question as to how to handle the iterative nature of property tax calculation in your analysis. There are a couple of options.

The first is to use Excel’s iterative calculation feature (File>Options>Formulas>Enable Iterative Calculation) and write a circular reference. But as I discuss in the linked thread above, I don’t recommend that.

The second is to do a side calculation to determined assessed value. This is what I do in practice as it’s the closest to what actually happens in the real world. Also, I prefer this method as it makes me think through the tax assessment rather than purely relying on the model’s built in calculation.

The third is to use the previous year’s NOI to calculate the tax assessment in the current year. I haven’t used this method yet in any of my models, but if you’re looking to make the property tax calculation fully dynamic, this would do it. And since property tax assessments lag the real value, basing the assessment off the previous year’s valuation makes sense. The only downside I see is that implementing this in a DCF would require tracking the property value at the end of each year.

Thanks again for the question, and please let me know if you have any others.

Spencer

[Spencer Burton]

Hi Phillip,

No stupid questions – especially when it relates to a complex topic like modeling real estate partnership structures!

To answer your question I thought I’d record a quick video that talks through why the logic in the All-in-One assumes the sponsor fees are positive.

 

Let me know if you have trouble viewing the video. Otherwise, happy to answer any follow up questions.

Spencer

[Spencer Burton]

You are correct – thanks for spotting the error!

I’ve gone ahead and updated the ‘Frost and Main GP, LLC summary partnership structure’ PDF to reflect a 40% promote to A.CRE Development Partners.

Sorry for the mistake!

[Spencer Burton]

Let me add a quick comment.

In practice, the prepayment penalty is generally calculated by the lender for the borrower. So the borrower most often requests a payoff, including the the prepayment penalty, and then the borrower uses that amount when modeling the transaction.

When doing back of the back of the envelope analysis, real estate professionals will generally just use an online defeasance or yield maintenance calculator to calculate the payoff and then plug that into their model.

In fact, I’ve had to model a deal with a defeasance just once in my career, and we used a defeancance calculator from a defeasance consultant! That calculator is here: https://www.defeasewithease.com/tools/defeasance-calculator

Nevertheless, I think it’s a good skill to know, only to better understand the concepts. So I’ll be sure to add a tutorial for calculating both at some point soon.

[Spencer Burton]

Kyle,

I hadn’t considered including a section on modeling defeasance or yield maintenance, but this is a great suggestion. I’ll plan to add a tutorial on the subject at some point in the near future. And as an initial launch member of the Accelerator, you have lifetime access to any changes to the program so I’ll be sure to let you know when I’ve added that tutorial.

Thanks for the suggestion!

Spencer

[Author]

Posts