Calculating Jarque Bera Statistic In Excel

Jarque-Bera Statistic Calculator for Excel

Calculate the Jarque-Bera test for normality using your Excel data. Enter your dataset below to compute skewness, kurtosis, and the Jarque-Bera statistic.

Comprehensive Guide: Calculating Jarque-Bera Statistic in Excel

The Jarque-Bera test is a powerful statistical tool used to determine whether sample data has the skewness and kurtosis matching a normal distribution. This guide will walk you through the complete process of calculating the Jarque-Bera statistic in Excel, interpreting the results, and understanding its practical applications in financial modeling, risk management, and data analysis.

Understanding the Jarque-Bera Test

The Jarque-Bera test is based on two moments of the data distribution:

  1. Skewness – Measures the asymmetry of the data around the sample mean
  2. Kurtosis – Measures the “tailedness” of the data distribution

The test statistic is calculated as:

JB = n/6 × (S² + (K-3)²/4)

Where:

  • n = sample size
  • S = sample skewness
  • K = sample kurtosis

Step-by-Step Calculation in Excel

Step 1: Prepare Your Data

Organize your data in a single column in Excel. Ensure there are no empty cells or non-numeric values in your dataset.

Example: Place your data in cells A1:A100

Step 2: Calculate Basic Statistics

Use these Excel functions:

  • =COUNT(A1:A100) for sample size (n)
  • =AVERAGE(A1:A100) for mean
  • =STDEV.P(A1:A100) for standard deviation

Step 3: Calculate Skewness

Excel’s built-in skewness function:

=SKEW(A1:A100)

Note: This calculates the population skewness. For sample skewness, you may need to adjust the formula.

Step 4: Calculate Kurtosis

Use Excel’s kurtosis function:

=KURT(A1:A100)

Note: Excel calculates excess kurtosis (subtracts 3 from the result).

Step 5: Compute Jarque-Bera Statistic

Combine the results using the formula:

=COUNT(A1:A100)/6*(SKEW(A1:A100)^2 + (KURT(A1:A100)^2)/4)

Step 6: Determine p-value

Use the chi-square distribution:

=CHISQ.DIST.RT(JB_statistic, 2)

Where JB_statistic is your calculated value from Step 5

Interpreting the Results

The Jarque-Bera test has two degrees of freedom (for skewness and kurtosis). The null hypothesis is that the data is normally distributed. Interpretation rules:

p-value Decision Interpretation
p > 0.05 Fail to reject H₀ Data appears normally distributed
p ≤ 0.05 Reject H₀ Data does NOT appear normally distributed
p ≤ 0.01 Strongly reject H₀ Strong evidence against normality

Practical Example with Financial Data

Let’s examine daily returns for S&P 500 index (2010-2020) with 2,517 observations:

Statistic Value Interpretation
Mean 0.00042 Near zero as expected for returns
Standard Deviation 0.0102 1.02% daily volatility
Skewness -0.382 Negative skew (more extreme negative returns)
Kurtosis 5.12 Fat tails (leptokurtic distribution)
Jarque-Bera 284.67 Very high value
p-value < 0.0001 Strongly reject normality

This analysis shows that financial returns typically deviate from normality, exhibiting negative skewness and fat tails – a phenomenon known as “stylized facts” in financial econometrics.

Common Mistakes to Avoid

  1. Small sample sizes: The Jarque-Bera test performs poorly with n < 50. For small samples, consider Shapiro-Wilk test instead.
  2. Outliers: Extreme values can disproportionately affect skewness and kurtosis calculations. Consider winsorizing your data.
  3. Excel version differences: Older Excel versions (pre-2010) may calculate kurtosis differently. Always verify your Excel version’s documentation.
  4. Population vs sample: Excel’s SKEW() and KURT() functions calculate population moments. For sample data, you may need to adjust the formulas.
  5. Non-numeric data: Ensure your dataset contains only numeric values. Text or empty cells will cause errors.

Advanced Applications

The Jarque-Bera test has numerous applications in finance and economics:

Asset Pricing Models

Testing normality of asset returns is crucial for:

  • Capital Asset Pricing Model (CAPM)
  • Black-Scholes option pricing
  • Value at Risk (VaR) calculations

Risk Management

Non-normal distributions affect:

  • Credit risk models
  • Operational risk measurements
  • Stress testing scenarios

Econometric Modeling

Normality assumptions underlie:

  • Ordinary Least Squares (OLS) regression
  • Maximum Likelihood Estimation (MLE)
  • Hypothesis testing procedures

Alternative Tests for Normality

While the Jarque-Bera test is powerful, consider these alternatives depending on your specific needs:

Test Best For Limitations Excel Implementation
Shapiro-Wilk Small samples (n < 50) Not available natively in Excel Requires VBA or add-ins
Anderson-Darling General purpose Complex calculation Requires specialized software
Kolmogorov-Smirnov Comparing with any distribution Less powerful for normality Not natively available
Lilliefors Modification of K-S for normality Not widely available Requires custom implementation
D’Agostino-Pearson Alternative to J-B Similar to J-B but different weights Complex Excel implementation

Excel VBA Implementation

For advanced users, here’s a basic VBA function to calculate the Jarque-Bera statistic:

Function JarqueBera(rng As Range) As Double
    Dim n As Long, i As Long
    Dim sum As Double, sum2 As Double, sum3 As Double, sum4 As Double
    Dim mean As Double, var As Double, skew As Double, kurt As Double
    Dim jb As Double

    n = Application.WorksheetFunction.Count(rng)
    sum = 0: sum2 = 0: sum3 = 0: sum4 = 0

    ' Calculate moments
    For i = 1 To n
        sum = sum + rng.Cells(i, 1).Value
        sum2 = sum2 + rng.Cells(i, 1).Value ^ 2
        sum3 = sum3 + rng.Cells(i, 1).Value ^ 3
        sum4 = sum4 + rng.Cells(i, 1).Value ^ 4
    Next i

    mean = sum / n
    var = sum2 / n - mean ^ 2
    skew = (sum3 / n - 3 * mean * var - mean ^ 3) / (var ^ (3 / 2))
    kurt = (sum4 / n - 4 * mean * (sum3 / n) + 6 * mean ^ 2 * var + 3 * mean ^ 4) / (var ^ 2) - 3

    ' Calculate Jarque-Bera statistic
    jb = n / 6 * (skew ^ 2 + kurt ^ 2 / 4)

    JarqueBera = jb
End Function
        

To use this function:

  1. Press Alt+F11 to open VBA editor
  2. Insert a new module (Insert > Module)
  3. Paste the code above
  4. In Excel, use =JarqueBera(A1:A100) where A1:A100 contains your data

Academic References and Further Reading

For a deeper understanding of the Jarque-Bera test and its applications:

  1. Jarque, C. M., & Bera, A. K. (1980). Efficient tests for normality, homoscedasticity and serial independence of regression residuals. Economics Letters, 6(3), 255-259 – The original paper introducing the test
  2. Federal Reserve Board (2012). Testing for Normality with GARCH Models – Application in financial econometrics
  3. William H. Greene (2018). Econometric Analysis, 8th Edition. Pearson – Comprehensive treatment in Chapter 17

Frequently Asked Questions

Q: What sample size is appropriate for the Jarque-Bera test?

A: The test works best with sample sizes greater than 50. For smaller samples (n < 50), consider the Shapiro-Wilk test instead, though it's not natively available in Excel.

Q: Why does my Jarque-Bera statistic differ from other software?

A: Differences can arise from:

  • Different skewness/kurtosis calculation methods (population vs sample)
  • Handling of missing values
  • Numerical precision differences
  • Whether excess kurtosis (kurtosis-3) is used

Q: Can I use this test for time series data?

A: Yes, but be cautious. The Jarque-Bera test assumes i.i.d. (independent and identically distributed) data. For time series with autocorrelation, you may need to:

  • Use returns instead of prices
  • Test residuals from an ARMA/GARCH model
  • Consider tests that account for serial correlation

Q: What should I do if my data fails the normality test?

A: Consider these approaches:

  • Transformations: Log, square root, or Box-Cox transformations
  • Non-parametric methods: Use tests that don’t assume normality
  • Robust estimators: Use median instead of mean, IQR instead of standard deviation
  • Model the non-normality: Use distributions like Student’s t, skewed t, or stable distributions
  • Bootstrap methods: Resampling techniques that don’t rely on distributional assumptions

Leave a Reply

Your email address will not be published. Required fields are marked *