xcorr — Cross-correlogram for bivariate time series

Description Quick start Menu Syntax

Options Remarks and examples Methods and formulas References
Also see

Description
xcorr plots the sample cross-correlation function.

Quick start
Plot sample cross-correlation function for y1 and y2 using tsset data
xcorr y1 y2
Same as above, but display a table instead of a graph
xcorr y1 y2, table
Same as above, but only include 10 lags and leads of the cross-correlation function
xcorr y1 y2, table lags(10)
Same as above, and generate xcorrvar containing the cross-correlation values
xcorr y1 y2, table lags(10) generate(xcorrvar)

Menu
Statistics > Time series > Graphs > Cross-correlogram for bivariate time series

1
 2 xcorr — Cross-correlogram for bivariate time series

Syntax
xcorr varname1 varname2 [ if ] [ in ] [ , options ]

options Description
Main
generate(newvar) create newvar containing cross-correlation values
table display a table instead of graphical output
noplot do not include the character-based plot in tabular output
lags(#) include # lags and leads in graph
Plot
base(#) value to drop to; default is 0
marker options change look of markers (color, size, etc.)
marker label options add marker labels; change look or position
line options change look of dropped lines
Add plots
addplot(plot ) add other plots to the generated graph
Y axis, X axis, Titles, Legend, Overall
twoway options any options other than by() documented in [G-3] twoway options
You must tsset your data before using xcorr; see [TS] tsset.
varname1 and varname2 may contain time-series operators; see [U] 11.4.4 Time-series varlists.

Options

 Main

generate(newvar) specifies a new variable to contain the cross-correlation values.

table requests that the results be presented as a table rather than the default graph.
noplot requests that the table not include the character-based plot of the cross-correlations.
lags(#) indicates the number of lags and leads to include in the graph. The default is to use
min(⌊ 𝑛/2 ⌋ − 2, 20).

 Plot

base(#) specifies the value from which the lines should extend. The default is base(0).
marker options, marker label options, and line options affect the rendition of the plotted cross-
correlations.
marker options specify the look of markers. This look includes the marker symbol, the marker size,
and its color and outline; see [G-3] marker options.
marker label options specify if and how the markers are to be labeled; see
[G-3] marker label options.
line options specify the look of the dropped lines, including pattern, width, and color; see
[G-3] line options.
 xcorr — Cross-correlogram for bivariate time series 3


 Add plots

addplot(plot) provides a way to add other plots to the generated graph; see [G-3] addplot option.

 Y axis, X axis, Titles, Legend, Overall

twoway options are any of the options documented in [G-3] twoway options, excluding by(). These
include options for titling the graph (see [G-3] title options) and for saving the graph to disk (see
[G-3] saving option).
stata.com

Remarks and examples

Example 1
We have a bivariate time series (Box et al. 2016 , Series J) on the input and output of a gas furnace,
where 296 paired observations on the input (gas rate) and output (% CO2 ) were recorded every 9 seconds.
The cross-correlation function is given by
. use https://www.stata-press.com/data/r18/furnace
(TIMESLAB: Gas furnace)
. xcorr input output, xline(5) lags(40)

Cross-correlogram
1.00 1.00
Cross-correlations of input and output

0.50 0.50

0.00 0.00

-0.50 -0.50

-1.00 -1.00
-40 -20 0 20 40
Lag
4 xcorr — Cross-correlogram for bivariate time series

We included a vertical line at lag 5, because there is a well-defined peak at this value. This peak indicates
that the output lags the input by five periods. Further, the fact that the correlations are negative indicates
that as input (coded gas rate) is increased, output (% CO2 ) decreases.
We may obtain the table of autocorrelations and the character-based plot of the cross-correlations
(analogous to the univariate time-series command corrgram) by specifying the table option.
. xcorr input output, table
-1 0 1
LAG CORR [Cross-correlation]

-20 -0.1033
-19 -0.1027
-18 -0.0998
-17 -0.0932
-16 -0.0832
-15 -0.0727
-14 -0.0660
-13 -0.0662
-12 -0.0751
-11 -0.0927
-10 -0.1180
-9 -0.1484
-8 -0.1793
-7 -0.2059
-6 -0.2266
-5 -0.2429
-4 -0.2604
-3 -0.2865
-2 -0.3287
-1 -0.3936
0 -0.4845
1 -0.5985
2 -0.7251
3 -0.8429
4 -0.9246
5 -0.9503
6 -0.9146
7 -0.8294
8 -0.7166
9 -0.5998
10 -0.4952
11 -0.4107
12 -0.3479
13 -0.3049
14 -0.2779
15 -0.2632
16 -0.2548
17 -0.2463
18 -0.2332
19 -0.2135
20 -0.1869

Once again, the well-defined peak is apparent in the plot.

 xcorr — Cross-correlogram for bivariate time series 5

Methods and formulas

The cross-covariance function of lag 𝑘 for time series 𝑥1 and 𝑥2 is given by

Cov{𝑥1 (𝑡), 𝑥2 (𝑡 + 𝑘)} = 𝑅12 (𝑘)

This function is not symmetric about lag zero; that is,

𝑅12 (𝑘) ≠ 𝑅12 (−𝑘)

We define the cross-correlation function as

𝑅𝑖𝑗 (𝑘)
𝜌𝑖𝑗 (𝑘) = Corr{𝑥𝑖 (𝑡), 𝑥𝑗 (𝑡 + 𝑘)} =
√𝑅𝑖𝑖 (0)𝑅𝑗𝑗 (0)

where 𝜌11 and 𝜌22 are the autocorrelation functions for 𝑥1 and 𝑥2 , respectively. The sequence 𝜌12 (𝑘) is
the cross-correlation function and is drawn for lags 𝑘 ∈ (−𝑄, −𝑄 + 1, . . . , −1, 0, 1, . . . , 𝑄 − 1, 𝑄).
If 𝜌12 (𝑘) = 0 for all lags, 𝑥1 and 𝑥2 are not cross-correlated.

References
Box, G. E. P., G. M. Jenkins, G. C. Reinsel, and G. M. Ljung. 2016. Time Series Analysis: Forecasting and Control. 5th
ed. Hoboken, NJ: Wiley.
Hamilton, J. D. 1994. Time Series Analysis. Princeton, NJ: Princeton University Press. https://doi.org/10.2307/j.
ctv14jx6sm.

Also see
[TS] corrgram — Tabulate and graph autocorrelations
[TS] pergram — Periodogram
[TS] tsset — Declare data to be time-series data

Stata, Stata Press, and Mata are registered trademarks of StataCorp LLC. Stata and Stata
®
Press are registered trademarks with the World Intellectual Property Organization of the
United Nations. StataNow and NetCourseNow are trademarks of StataCorp LLC. Other
brand and product names are registered trademarks or trademarks of their respective com-
panies. Copyright © 1985–2023 StataCorp LLC, College Station, TX, USA. All rights
reserved.
For suggested citations, see the FAQ on citing Stata documentation.