Skip to content
All library documents

Approximating Compound Interest with the Taylor Series for Logarithms

Article Quant Q&A · Author: thisisme

Summary

The document explains how to approximate the value of repeatedly compounded interest by rewriting the power expression with an exponential and a logarithm. It uses the Taylor series for the logarithm of one plus the periodic rate, then substitutes successive truncated sums into the exponential. The first-order approximation keeps only the rate term, while the second-order version adds a correction for the squared rate.

A worked example starts with a stated principal, annual rate, and investment horizon. The document compares the exact compounded amount with approximations using increasing numbers of series terms, showing closer agreement as more terms are included for that example. The series converges for absolute rates below one, while a truncated approximation is not exact and its accuracy depends on the rate and number of terms retained. The calculation teaches a general approximation technique, though it is a mathematical illustration rather than a trading strategy or market analysis.

Key ideas

  • Compounded growth can be rewritten as an exponential involving the logarithm of one plus the periodic rate.
  • The logarithm's Taylor series yields successively more accurate approximations when more terms are retained.
  • The second-order formula accounts for the squared-rate correction.
  • The worked example compares truncated approximations with the exact compounded value.
  • The logarithmic series converges when the absolute value of the rate is below one.

Tags

Full text
# taylor expansion in compounded interest


# taylor expansion in compounded interest












You invest $1, 000$ dollars for $10$ years at a $5$% yearly interest rate. After each year the interest paid is reinvested at the same rate.

(a) Represent the total amount A after ten years in the form $A = S(1 + x)^n$. What are $S$, $x$, and $n$ here?

$ S = 1000, x = .05, n =10$

(b) Compute A (up to cents) using a calculator.

Plugging in the corresponding numbers

$A=1628.89$ dollars

This is where I begin to struggle with the problem

(c) If one writes $A$ in the form $A = S(1+x)^n = S\exp(n\log(1+x))$ and uses the approximation $log(1 + x) ≈ x$ for small $x$, then one obtains $A ≈ Se^{nx}$. Which value for A is obtained with this approximation method?

(d) A better approximation is $log(1 + x) ≈ x − 1/2 x^2$. Explain how this is related to finding the taylor approximation for $\log(1+x)$ which is

$log(1+x)=x-x^2/2+x^3/3-x^4/4+x^5/5+...=\sum_{n=1}^\infty (-1)^{n+1}\frac{x^n}{n}$

(e) Based on the approximation in (d) develop an approximation formula for A in terms of S, x, and n. Use it to compute A in our specific example.

## Answer by caverac (score 2, accepted)

https://quant.stackexchange.com/a/36269

> (c) If one writes $A$ in the form $A=S(1+x)^n=S e^{n\ln(1+x)}$ and uses the approximation $\ln(1+x)\approx x$ for small $x$, then one obtains $A\approx S e^{nx}$. Which value for $A$ is obtained with this approximation method?

This is is based on the properties

\begin{eqnarray} \ln (a^b) &=& b\ln a \tag{1a}\\ e^{\ln a} &=&a \tag{1b} \end{eqnarray}

Therefore

$$ A = S(1+x)^n \stackrel{(1b)}{=} Se^{\ln(1 + x)^n} \stackrel{(1a)}{=} Se^{n\ln(1 + x)} \tag{2} $$

Now, you can use the fact that when $|x|< 1$ then

$$ \frac{1}{1+x} = 1 - x + x^2 - \cdots = \sum_{k=0}^{+\infty}(-1)^k x^k $$

if you integrate both sides, you get

$$ \ln(1 + x) = x -\frac{x^2}{2} + \frac{x^3}{3}- \cdots = \sum_{k=0}^{\infty}(-1)^k \frac{x^{k+1}}{k+1} \tag{3} $$

such that for small $x$ it is enough to take just the first term in this sum

$$ \ln(1 + x)\approx x \tag{4} $$

Eqn (2) then becomes

$$ A_1\approx Se^{nx} = 1648.72 $$

I will use the subscript $1$ to emphasize we are taking just the first term in the series

> (d) A better approximation is $\ln(1 + x) \approx x - x^2/2$. Explain how this is related to finding the Taylor approximation of $\ln(1 + x)$ which is $$ \ln(1 + x) = x -\frac{x^2}{2} + \frac{x^3}{3}- \cdots = \sum_{k=0}^{\infty}(-1)^k \frac{x^{k+1}}{k+1} $$

You just need to take the first two terms in Eq. (3). Here is the thing, if $x$ is small $x^2/2$ is smaller, and $x^3/3$ is even smaller, and $\dots$ So the more terms you include the better, but the smaller the correction in the overall result. Up to second order

$$ A_2\approx Se^{n(x - x^2/2)} = 1628.24 $$

which looks almost like the exact result you found before. If you're to include up to third order you'd see how it approaches even better the exact solution.

> e) Based on the approximation in (d) develop an approximation formula for $A$ in terms of $S$, $x$, and $n$. Use it to compute $A$ in our specific example.

In general if you take up to $N$ terms in the sum this is what you get

$$ A_N \approx Se^{n(x - x^2 - \cdots + (-1)^{N+1}x^{N}/N)} $$

Below there's some results

\begin{eqnarray} A_{\rm exact} &=& 1628.89 \\ A_1 &=& 1648.72 \\ A_2 &=& 1628.24 \\ A_3 &=& 1628.92 \\ A_4 &=& 1628.89 \end{eqnarray}

Shown in full with attribution under the source's licence. Licence: CC BY-SA 4.0 (Stack Exchange)

This summary was written by Stratmill's research agent from the original; it is not a copy of the source.