It is fairly well documented that investment bankers aspiring to exit into private equity have to do an Excel test/financial modeling test as part of the interview.
We confirmed that this is generally the case for interviews at any reputable PE firm – and it is also the case when investment banking analysts or associates are trying to lateral over to an elite boutique such as Evercore.
Now the modeling test is just one component of the interview – and everyone is expected to ace it. Building a good LBO in a compressed time period and coming up with and articulating the right conclusions is the bare minimum, so anyone looking to get into some of these roles should be able to do it like clockwork.
Financial Modeling Test for Investment Banking
As we alluded to in a previous post, the exam is almost always a leveraged buyout case study.
Sometimes there is a discounted cash flow valuation as well, and the DCF and LBO can be sense checks for each other to see if the value the interview candidate triangulates to makes sense.
The simplest LBO model will just consist of the interview candidate being given an EBITDA line, a purchase multiple, exit multiple, initial leverage and some other assumptions (interest rate, inflation/growth).
This is usually for less reputable private equity firms because candidates can basically memorize the model infrastructure and churn it out without understanding corporate finance deeply.
Now 90% of the time, the modeling test actually goes as follows.
Components of the Private Equity Modeling Test
So for the respectable version of this modeling test, the interviewer will usually give the candidate a public information binder/PIB for a company that is not a big headliner (such as Apple or Google) that is dated (let’s say 2010, but you will find that investment banking has barely changed over the years).
So from the last year’s financials (adjusting for one quarter if the interviewer wants to be extra annoying) and equity research from the time period, candidates have to construct a leveraged buyout and answer qualitative questions around whether or not the company makes sense for a private equity firm to purchase.
The candidate will also have to have some reasonable assumptions around how much leverage they can put on, because this will not be given. So it makes sense to ask the leveraged finance team at your investment bank on what creditors would accept or hop on S&P’s leveraged loan credit analytics website. This means what leverage (debt/EBITDA) the company can support and what the going interest rate would be at that leverage for a company in that industry.
The equity research itself should be ignored other than to grab the share price/current valuation in most cases. This is not a hedge fund interview where you are trying to develop a thesis on the business – this is a test of corporate finance math unless instructed otherwise.
Basically, equity research and current trading multiples determine what you would expect to pay – obviously a premium to that multiple – and that informs the internal rate of return that dictates the LBO or no LBO decision.
Building the LBO for the Modeling Test
Usually, better private equity firms require candidates to start from a blank Excel workbook – this tells them which analysts can operate without being shielded by templates.
I usually like to start by building the model infrastructure first and keeping a separate assumptions tab.
Then I will take these steps.
1. Build skeleton of income statement ending with net income
2. Build skeleton of cash flow statement starting with net income from income statement
3. Build basic net debt schedule with revolver and separate cash balance
4. Build IRR calculation and equity schedule based on purchase multiple at beginning and exit multiple at end
5. Populate an assumptions or input tab to feed the model
6. Build a circuit breaker (actually do this step earlier or when things #REF out or #VALUE
If a full set of financials are presented, there may be expectations for a working capital schedule. So accounts receivables should be a percentage of sales and inventory and accounts payable should be a function of cost of goods sold. Obviously, if sales are increasing and COGS are increasing, net working capital should be increasing and lowering free cash flow. Make sure that this dynamic is at play – if the number signs are switched the other way you will be in trouble.
Again, learning how to build a working revolver is key. If there is positive free cash flow, the revolver should be paid down. If there is cash before the revolver is paid down, the excess cash should be used to pay down the balance unless there is a minimum cash balance provision. Candidates should test the revolver by jamming a negative number into free cash flow to see if it works properly.
The Private Equity Case Study
Now these models can be much more advanced because the interviewer may expect bullet debt (bonds, TLB) as well as the revolving debt.
Or they can just ask qualitatively after the modeling test to see if the candidate understands how different tranches of capital would affect IRR.
So for example they could say – how would the IRR change if we implemented:
1. Cash Sweep
2. Tranche of subordinated debt
3. Preferred equity
Obviously, higher cost of capital tranches will lower the IRR as they eat into returns for equityholders.
After the test, there can be a write up or a panel interview asking whether or not the investment makes sense under the current assumptions. If the purchase multiple is high, this usually means a lower IRR – and this may mean no. The candidate will be expected to defend their assumptions and be asked questions on how the investment thesis may change if X or Y happens.
The difference in difficulty comes in the form of a time constraint. So some pension funds and private equity firms such as CPPIB and certain groups at Brookfield may give 4 hours to conduct the test – this is an extremely generous time frame and anyone who understands corporate finance theory can do this.
However, KKR and Blackstone may only give a candidate 2 hours. This is more of a real test with time pressure that will affect even good modelers. This means that candidates that like to overbuild models will struggle – every line in the model should be something that drives the model.