using financial analysis techniques for non-profit fundraising
DESCRIPTION
From a talk given at the 2013 APRA Symposium on Data Analytics in Baltimore, MD.TRANSCRIPT
Using Financial Analysis Techniques
in non-profit fundraising
Data Analytics SymposiumAugust 7 – 8, 2013 • Baltimore, Maryland
By Tommy TavennerNational Wildlife Federation
Data Analytics SymposiumAugust 7-8, 2013
Financial Engineering
• Application of theoretical mathematics and computer science to the study of financial data
• The basis for many of our analysis applications
• Uses the same basic mathematics techniques and requires the same cautions
© 2013 Tommy Tavenner
Data Analytics SymposiumAugust 7-8, 2013
• A different way of thinking about our donors
• The balance of risk vs. reward
• Focus on minimizing risk, defined as volatility
• Finding the right mix of donors for the outcome we are trying to achieve
Why you should study it
© 2013 Tommy Tavenner
Data Analytics SymposiumAugust 7-8, 2013
• Prospect Research = Fundamental Analysis
• Analytics = Technical Analysis
How It Relates
© 2013 Tommy Tavenner
Data Analytics SymposiumAugust 7-8, 2013
Fundamental Analysis
• Determine the value of a company by looking at it’s principal components
> Financial Statements
> Management Team
> Competitive Environment
© 2013 Tommy Tavenner
Data Analytics SymposiumAugust 7-8, 2013
Technical Analysis
• Based on the idea that all information about a company is contained in it’s share price
> Location of a company share price within the larger market
> Future price
> Relationship of movements within prices of multiple companies
© 2013 Tommy Tavenner
Data Analytics SymposiumAugust 7-8, 2013
Some caveats
• Our information is less complete
• The amount we invest in a donor is more highly correlated with their giving
© 2013 Tommy Tavenner
Data Analytics SymposiumAugust 7-8, 2013
Definitions
• Asset/Security – an individual stock or fund; equivalent to the donor
• Alpha – A measure of the risk adjusted return
• Beta – A measure of risk; The volatility of an asset (i.e. how much the price swings) in relation to a benchmark (i.e. the overall market)
© 2013 Tommy Tavenner
Data Analytics SymposiumAugust 7-8, 2013
Concepts
• Market Rate
> Historical – The average gift for your overall population in each measured period, i.e. average per year, per quarter, per month, etc.
> Estimated – A forecast of what the average gift will be in the next measured period.
© 2013 Tommy Tavenner
Data Analytics SymposiumAugust 7-8, 2013
Concepts
• Risk Free Rate
> The safest investment you can make with a nearly 100% guarantee return. (treasury bonds)
> Low risk generally means low returns as well.
> What size gift would we receive if we made little to no effort at all
1. Unsolicited/White mail gifts
2. Base level giving or membership fee
© 2013 Tommy Tavenner
Data Analytics SymposiumAugust 7-8, 2013
Concepts
• Calculating Beta for a donor
Donor’s Gifts per Year Average Giving per Year
$ 100 $ 80
$120 $ 94
$120 $ 90
$115 $ 93
β = CoVar( Donor’s Gifts, Average Giving )
Var( Average Giving )
© 2013 Tommy Tavenner
Data Analytics SymposiumAugust 7-8, 2013
Concepts
• Calculating Beta for a donor
β = CoVar
=Var
$ 100 $ 80
$120 $ 94
$120 $ 90
$115 $ 93
$ 80
$ 94
$ 90
$ 93
55.417
40.917= 1.354
© 2013 Tommy Tavenner
Data Analytics SymposiumAugust 7-8, 2013
Concepts - Beta
β < 0 The size of the donor’s gift moves inverse to the overall population
β = 0 The size of the donor’s gift is uncorrelated to the population
0 < β < 1 The size of the donor’s gift moves in the same direction but not as dramatically as the overall population
β = 1 The size of the donor’s gift is exactly correlated thepopulation
β > 1 The size of the donor’s gift moves in the same direction but more dramatically as the population
© 2013 Tommy Tavenner
Data Analytics SymposiumAugust 7-8, 2013
CAPITAL ASSETS PRICING MODEL
Data Analytics SymposiumAugust 7-8, 2013
Capital Assets Pricing Model
• Based on the work of Harry Markowitz
• Introduced by Jack Treynor, William Sharpe, John Lintner, and Jan Mossin independently over the course of several studies between 1961 and 1966
• A method for determining an assets required rate of return given its inherent risk
© 2013 Tommy Tavenner
Data Analytics SymposiumAugust 7-8, 2013
Capital Assets Pricing Model
• A method for determining an assets required rate of return given its inherent risk
> Requires knowing a benchmark rate of return and a risk free rate of return
> Also depends on the asset’s beta (β), a measure of the volatility of that asset
© 2013 Tommy Tavenner
Data Analytics SymposiumAugust 7-8, 2013
Capital Assets Pricing Model
• Expected Market Rate
> Average Return for all donors
> Average Return for the specific population being studied
> Major Donors: conversion rate (gifts over asks or dollars received over dollars asked)
© 2013 Tommy Tavenner
Data Analytics SymposiumAugust 7-8, 2013
Capital Assets Pricing Model
• Risk Free Return Rate = What activities have set returns?
> Treasury Bonds
> Benchmarking Studies (i.e. CASE, Blackbaud)
> Donors with close to 100% response rate
> White mail
© 2013 Tommy Tavenner
Data Analytics SymposiumAugust 7-8, 2013
Capital Assets Pricing Model – An example
• Annual Giving – Gifts of $1,000 or more
> Set a fixed time for your study. Requires a minimum of two values (i.e. the first of two consecutive years, but more is generally better
1. Past performance of the target population
2. Past performance of your benchmark
3. Estimate of benchmark performance in your prediction period
4. Risk Free Rate
© 2013 Tommy Tavenner
Data Analytics SymposiumAugust 7-8, 2013
Capital Assets Pricing Model – An example
• Annual Giving – Gifts of $1,000 or more
> Study period: The past four years
1. Past performance of the target population: Giving per individual by year
2. Past performance of your benchmark: Average giving by year
3. Estimate of benchmark performance in your prediction period:
— Excel FORECAST function
4. Risk Free Rate: Base giving level - $1,000
© 2013 Tommy Tavenner
Data Analytics SymposiumAugust 7-8, 2013
Capital Assets Pricing Model – An example
= $1,000 + 1.08( $1,030 – $1,000 )
= $1,032.26
Risk Free Rate: $1,000
Market Forecast: $1,030
Donor 1 β: 1.08
Estimated Return for Donor 1
© 2013 Tommy Tavenner
Data Analytics SymposiumAugust 7-8, 2013
Capital Assets Pricing Model – An example
= 0 + 1.08( $1,030 –0 )
= $1,107.46
Risk Free Rate: $0
Market Forecast: $1,030
Donor 1 β: 1.08
Estimated Return for Donor 1
© 2013 Tommy Tavenner
Data Analytics SymposiumAugust 7-8, 2013
Capital Assets Pricing Model – An example
β Giving Forecast CAPM Forecast($1,000 RFR)
CAPM Forecast($0 RFR)
Donor 1 1.08 $ 1,250 $ 1,032 $ 1,107
Donor 2 0.46 $ 775 $ 1,014 $ 472
Donor 3 1.67 $ 1,250 $ 1,050 $ 1,725
Donor 4 2.20 $ 750 $ 1,066 $ 2,269
Donor 5 -0.41 $ 1,125 $ 987 $ (423.62)
© 2013 Tommy Tavenner
Data Analytics SymposiumAugust 7-8, 2013
Modern Portfolio Theory
Data Analytics SymposiumAugust 7-8, 2013
Modern Portfolio Theory
• Attempts to maximize the return of a portfolio for a given amount of risk
• Focuses on diversification of assets based on their individual volatility.
© 2013 Tommy Tavenner
Data Analytics SymposiumAugust 7-8, 2013
Modern Portfolio Theory
• Requirements:
> an estimate of the overall portfolio’s return
> Understanding the correlation of the component assets
© 2013 Tommy Tavenner
Data Analytics SymposiumAugust 7-8, 2013
Modern Portfolio Theory
• Expected Return
> Many methods to forecast future returns
1. Linear/Non linear regression
2. Simple averaging
3. Seasonality and Time Series Analysis
> Depends on the giving patterns of your donors
© 2013 Tommy Tavenner
Data Analytics SymposiumAugust 7-8, 2013
Modern Portfolio Theory
• Portfolio Correlation
> Looking at the observations for each donor, how closely related is the movement of their giving.
> Example
Year 1 Year 2 Year 3 Year 4
Donor 1 $ 100 $ 120 $120 $ 115
Donor 2 $ 75 $ 100 $ 80 $ 75
Donor 3 $ 100 $ 125 $ 150 $ 125
Donor 4 $ 50 $ 75 $ 50 $ 50
Donor 5 $ 75 $ 100 $ 100 $125
© 2013 Tommy Tavenner
Data Analytics SymposiumAugust 7-8, 2013
Modern Portfolio TheoryPortfolio Correlation
$-
$20
$40
$60
$80
$100
$120
$140
$160
Year 1 Year 2 Year 3 Year 4
Donor 1
Donor 2
Donor 3
Donor 4
Donor 5
© 2013 Tommy Tavenner
Data Analytics SymposiumAugust 7-8, 2013
Modern Portfolio Theory
• Portfolio Correlation
Donor 1 Donor 2 Donor 3 Donor 4 Donor 5
Donor 1
Donor 2 55.48 %
Donor 3 86.27 % 17.15 %
Donor 4 44.02 % 98.02 % 0.00 %
Donor 5 64.70 % 0.00 % 50.00 % 0.00 %
© 2013 Tommy Tavenner
Data Analytics SymposiumAugust 7-8, 2013
Modern Portfolio Theory
• The Efficient Frontier
> Created by Harry Markowitz
> Finds the lowest risk portfolio for a target return
> Accomplishes this by minimizing standard deviation
© 2013 Tommy Tavenner
Data Analytics SymposiumAugust 7-8, 2013
Modern Portfolio Theory
• Covariance Matrix
Donor 1 Donor 2 Donor 3 Donor 4 Donor 5
Donor 167.1875 46.875 125 39.0625 93.75
Donor 246.875 106.25 31.25 109.375 0
Donor 3125 31.25 312.5 0 156.25
Donor 439.0625 109.375 0 117.1875 0
Donor 593.75 0 156.25 0 312.5
© 2013 Tommy Tavenner
Data Analytics SymposiumAugust 7-8, 2013
Modern Portfolio Theory
• The Efficient Frontier
Average Gift $ 31.25 $ 35.00 $ 40.00 $ 45.00 $ 50.00 $ 55.00
Standard Deviation 16.298 16.421 16.583 16.744 16.903 17.060
Slope 1.917 2.131 2.412 2.687 2.958 3.223
Donor 1 0% 0% 0% 0% 0% 0%
Donor 2 0% 14% 33% 52% 71% 90%
Donor 3 0% 0% 0% 0% 0% 0%
Donor 4 100% 86% 67% 48% 29% 10%
Donor 5 0% 0% 0% 0% 0% 0%
© 2013 Tommy Tavenner
Data Analytics SymposiumAugust 7-8, 2013
Modern Portfolio – The Efficient Frontier
$-
$20.00
$40.00
$60.00
$80.00
$100.00
© 2013 Tommy Tavenner
Data Analytics SymposiumAugust 7-8, 2013
Modern Portfolio Theory
• Minimizing Beta
> Calculate the beta of each donor
> Portfolio Beta = β₁ω₁ + β2ω2 + … + βnωn
> Option 1: Find top X donors who produce the lowest β
> Option 2: Find the appropriate weights for a given set of donors
that minimizes overall β
© 2013 Tommy Tavenner
Data Analytics SymposiumAugust 7-8, 2013
Other Measures to Explore
Data Analytics SymposiumAugust 7-8, 2013
Sharpe Ratio
• Measures the return of an asset per unit of deviation.
• How much reward for how much risk
© 2013 Tommy Tavenner
Data Analytics SymposiumAugust 7-8, 2013
Sharpe Ratio
• How it applies to fundraising
• Unlike β, the Sharpe Ratio does not compare a donor’s
volatility to that of a benchmark
• Useful for comparing the relative volatility of donors across different giving circles.
© 2013 Tommy Tavenner
Data Analytics SymposiumAugust 7-8, 2013
Sortino Ratio
• A modification of the Sharpe ratio that only penalizes downside volatility
• Requires a minimum acceptable return (MAR)
• Only values which fall below the MAR are counted in the formula
© 2013 Tommy Tavenner
Data Analytics SymposiumAugust 7-8, 2013
Sortino Ratio
• How it applies to fundraising
• Just like in finance, we are usually only concerned with volatility that causes giving to decrease.
© 2013 Tommy Tavenner
Data Analytics SymposiumAugust 7-8, 2013
Treynor Ratio
• Returns subtracted from the risk free rate and divided by beta
© 2013 Tommy Tavenner
Data Analytics SymposiumAugust 7-8, 2013
Treynor Ratio
• How it applies to fundraising
• More granular than Sharpe, you can see how a donor or marketing effort’s independent return (i.e. in addition to the benchmark) compares to its independent volatility.
© 2013 Tommy Tavenner
Data Analytics SymposiumAugust 7-8, 2013
Jensen’s Alpha
• Measures the additional return over a theoretical return.
• The theoretical return is typically the value derived from CAPM
© 2013 Tommy Tavenner
Data Analytics SymposiumAugust 7-8, 2013
Jensen’s Alpha
• How it applies to fundraising
• Use to compare performance against both your CAPM value and projected value to see how closely it performed.
© 2013 Tommy Tavenner