# Split equation with each element being exactly below each other

I have an estimated equation with both coefficients and t-statistics. It's a multiline equation and would like to put the t-statistics right below the coefficient estimates Currently my code looks like this:

$$\label{eq:test} \begin{split} EST =& 4.00 - 7.00BETA - 2.00SMB + 8.00HML + 3.00MOM + 1.00SMBU - \\ & (8.00) (-2.20) (-1.60) (9.00) (3.00) (1.00) \\ & 1.00TERMR - 1.00SMBH + 1.00SMBL + 0.00SMBK - 1.30HMLR.\\ & (-1.11) (1.53) (6.81) (2.14) (-1.85) \end{split}$$


So the 4.00 in the first line should be right above the (8.00) from the line below. The 7.00 in the first line should be exactly above the (-2.20). The 2.00 in the first line should be right above the (-1.60) and so on for the other coefficient and t-statistics, respectively. It should look like equation (1) from the Fama/French (1993) paper for those of you who are into finance:

-

There are many ways to achieve this. Here's one where a [t]op-aligned array is used to place the coefficient and stacked below with the t-statistic. This is all achieved via \coeft{<coef>}{<t-stat>}:

\documentclass{article}
\usepackage{mathtools}% http://ctan.org/pkg/amsmath
\newcommand{\coeft}[2]{
\begin{array}[t]{@{}c@{}}
#1\\[\dimexpr-\normalbaselineskip+1em]
\scriptstyle\mathclap{(#2)}
\end{array}}
\begin{document}

$$\label{eq:test1} \begin{split} EST =& 4.00 - 7.00BETA - 2.00SMB + 8.00HML + 3.00MOM + 1.00SMBU - \\ & (8.00) (-2.20) (-1.60) (9.00) (3.00) (1.00) \\ & 1.00TERMR - 1.00SMBH + 1.00SMBL + 0.00SMBK - 1.30HMLR.\\ & (-1.11) (1.53) (6.81) (2.14) (-1.85) \end{split}$$

$$\label{eq:test2} \begin{split} \text{EST} &= \coeft{4.00}{8.00} - \coeft{7.00}{-2.20}\text{BETA} - \coeft{2.00}{-1.60}\text{SMB} + \coeft{8.00}{9.00}\text{HML} + \coeft{3.00}{3.00}\text{MOM} + \coeft{1.00}{1.00}\text{SMBU} \\ & \phantom{=}{} - \coeft{1.00}{-1.11}\text{TERMR} - \coeft{1.00}{1.53}\text{SMBH} + \coeft{1.00}{6.81}\text{SMBL} + \coeft{0.00}{2.14}\text{SMBK} - \coeft{1.30}{-1.85}\text{HMLR.} \end{split}$$

\end{document}


Some changes to your code include using:

• \text{<var>} to represent the variable <var>. Otherwise, for say BETA, TeX will consider this as a product between the variables (letters) B, E, T and A;
• &= rather than =& for the alignment of the equation;
• spacing techniques such as \phantom{=} to obtain the correct indentation at alignment around =;
• {} before binary operators that have no counterpart in order to obtain the correct spacing (see the minus sign before 1.00TERMR, which would otherwise have been seen as a unary/negative sign;
• \scriptstyle to represent the t-statistics, to not emphasize these as being part of the equation.

mathtools provides \mathclap which is a centred overlap in math mode so that any of the t-statistics do not influence the equation spacing.

Of course, you can modify \coeft to your liking. I've included the brackets (), but this can be removed.

-
Worked great! Thank you very much! – ben Apr 25 '13 at 21:03
$$\label{eq:test} \begin{array}{*7{l}} EST =& 4.00& {}- 7.00BETA& {}- 2.00SMB&{} + 8.00HML& + 3.00MOM& + 1.00SMBU - \\ & (8.00)& (-2.20)& (-1.60)& (9.00)& (3.00)& (1.00) \\ & 1.00TERMR& - 1.00SMBH& + 1.00SMBL &+ 0.00SMBK &- 1.30HMLR.&\\ & (-1.11) &(1.53)& (6.81)& (2.14) &(-1.85)& \end{array}$$


This is a general idea. Some adjustments, especially {} groups (three of them added), are needed.

-