# Two tables underneath each other with three part table

I am trying to put two tables underneath each other within a threeparttable environment. My TeX code looks as follows:

\documentclass[11pt,a4paper]{article}
\usepackage{amssymb, amsmath, bm}
\usepackage[flushleft]{threeparttable}
\usepackage{float, booktabs, makecell, caption, tabularx}
\usepackage{siunitx}

\begin{document}

\begin{table}[htb!]
\caption{\textbf{Estimates of the MS-TVTP model}}
\label{table:mstvtp_estimation_results}
\begin{threeparttable}
\footnotesize
\renewcommand{\TPTminimum}{\linewidth}
\makebox[\linewidth]{%
\tabcolsep=0.11cm
\setlength{\extrarowheight}{2pt}
\begin{tabular}{lccc|ccccc}
\multicolumn{7}{l}{Panel (A): Autoregressive matrix $\boldsymbol{F}$, constants $\boldsymbol{\mu}_1$ and $\boldsymbol{\mu}_2$ and decay parameter $\lambda$} \\
\bottomrule
\specialrule{0.4pt}{\aboverulesep}{0pt}
& \thead[l]{$\mathrm{Level}_{t-1}$} & \thead[l]{$\mathrm{Slope}_{t-1}$} &
\thead[l]{$\mathrm{Curvature}_{t-1}$} & $\boldsymbol{\mu}_1$ & $\boldsymbol{\mu}_2$ & $\lambda$ & $p_{11}$ & $p_{22}$ \\
\Xhline{0.5pt}
$\mathrm{Level}_{t}$ $(\beta_{1,t})$ & $\bm{0.996}$ & & & 0.055 & & $\bm{0.039}$ & 0.928 & 0.922 \\
& (0.003) & & & (0.038) & & (0.001)  \\
$\mathrm{Slope}_{t}$ $(\beta_{2,t})$ & & & & $\bm{-1.412}$ &
$\bm{-4.936}$ \\
& & & & (0.075) & (0.106) \\
$\mathrm{Curvature}_{t}$ $(\beta_{3,t})$ & & & $\bm{0.979}$ & -0.088 & \\
& & & (0.010) & (0.046) & \\
\Xhline{0.8pt}
\end{tabular}
\\
\begin{tabular}{lcc}
\multicolumn{3}{l}{Panel (B): Time-varying transition probability matrix $P_t$} \\
\bottomrule
\specialrule{0.4pt}{\aboverulesep}{0pt}
&  \thead[l]{$S_t = 1$} & \thead[l]{$S_t = 2$} &  \\
\Xhline{0.5pt}
$\gamma_0$ & $\bm{7.000}$ & $\bm{-0.999}$ & & \\
& (0.112)  & (0.460)        &                 & &       &        \\
$\gamma_1$ & $\bm{1.834}$ & $\bm{-0.626}$ & &       &        \\
& (0.659)  & (0.277)        & &               &         &        \\
$\gamma_2$ & 0.420          & 0.369           & &       &        \\
& (0.602)  & (0.514)        & &               &         &        \\
$\gamma_2$ & -0.292         & $\bm{2.000}$  & &       &        \\
& (0.839)  & (0.597)        & &               &         &        \\
\Xhline{0.8pt}
\end{tabular}
}
\begin{tablenotes}[flushleft]\footnotesize\smallskip
\item Note: This table reports estimates of the MS-DNS model. Panel (A) presents estimates for the autoregressive coefficient matrix $\boldsymbol{F}$, vector of means during periods of normal interest rates ($\boldsymbol{\mu}_1$) and in a low interest rate environment ($\boldsymbol{\mu}_2$), and decay parameter $\lambda$. Panel (B) presents estimates of the covariance matrix $\boldsymbol{\Sigma}_{\eta}$, and the transition probabilities $p_{11}$ and $p_{22}$. Bold entries denote parameter estimates significant at the 5 percent level. Standard errors appear in parentheses.\par
\end{tablenotes}
\end{threeparttable}
\end{table}

\end{document}


Does anybody know how to do this? Also, I would like the second table (which is quite smaller than the first one) to be centered in the middle in case that does not happen automatically.

Edit: It is not required that the two tables should stay in the threeparttable environment.

• In your second table, you use 7 columns but nly specify 3. Also, is there a special reasen why you want to use threeparttable for this? – leandriis Jun 17 at 19:42
• No, there is no particular reason. I first had one big table in threeparttable but later on decided to seperate it into two smaller ones to make it visually more clear. – Matthew Jun 17 at 19:58

Without the threeparttable, the \makebox and the superfluous & in the second table, but with a smaller font size and some other rearrangements:

\documentclass[11pt,a4paper]{article}
\usepackage{amssymb, amsmath, bm}
\usepackage[flushleft]{threeparttable}
\usepackage{float, booktabs, makecell, caption, tabularx}
\usepackage{siunitx}
\usepackage{ragged2e}

\begin{document}

\begin{table}[htb!]

\caption{\textbf{Estimates of the MS-TVTP model}}
\label{table:mstvtp_estimation_results}

\centering\scriptsize
\renewcommand{\TPTminimum}{\linewidth}

\setlength{\tabcolsep}{2pt}
\setlength{\extrarowheight}{2pt}
\begin{tabular}{lccc|ccccc}
\multicolumn{7}{l}{Panel (A): Autoregressive matrix $\boldsymbol{F}$, constants $\boldsymbol{\mu}_1$ and $\boldsymbol{\mu}_2$ and decay parameter $\lambda$} \\
\bottomrule
\specialrule{0.4pt}{\aboverulesep}{0pt}
& \thead[l]{$\mathrm{Level}_{t-1}$} & \thead[l]{$\mathrm{Slope}_{t-1}$} &
\thead[l]{$\mathrm{Curvature}_{t-1}$} & $\boldsymbol{\mu}_1$ & $\boldsymbol{\mu}_2$ & $\lambda$ & $p_{11}$ & $p_{22}$ \\
\Xhline{0.5pt}
$\mathrm{Level}_{t}$ $(\beta_{1,t})$ & $\bm{0.996}$ & & & 0.055 & & $\bm{0.039}$ & 0.928 & 0.922 \\
& (0.003) & & & (0.038) & & (0.001)  \\
$\mathrm{Slope}_{t}$ $(\beta_{2,t})$ & & & & $\bm{-1.412}$ &
$\bm{-4.936}$ \\
& & & & (0.075) & (0.106) \\
$\mathrm{Curvature}_{t}$ $(\beta_{3,t})$ & & & $\bm{0.979}$ & -0.088 & \\
& & & (0.010) & (0.046) & \\
\Xhline{0.8pt}
\end{tabular}

\bigskip

Panel (B): Time-varying transition probability matrix $P_t$

\begin{tabular}{lcc}
\bottomrule
\specialrule{0.4pt}{\aboverulesep}{0pt}
&  \thead[l]{$S_t = 1$} & \thead[l]{$S_t = 2$} \\
\Xhline{0.5pt}
$\gamma_0$ & $\bm{7.000}$ & $\bm{-0.999}$  \\
& (0.112)  & (0.460)                           \\
$\gamma_1$ & $\bm{1.834}$ & $\bm{-0.626}$  \\
& (0.659)  & (0.277)                           \\
$\gamma_2$ & 0.420          & 0.369            \\
& (0.602)  & (0.514)                           \\
$\gamma_2$ & -0.292         & $\bm{2.000}$   \\
& (0.839)  & (0.597)                           \\
\Xhline{0.8pt}
\end{tabular}
\medskip

\justifying\footnotesize\noindent
Note: This table reports estimates of the MS-DNS model. Panel (A) presents estimates for the autoregressive coefficient matrix $\boldsymbol{F}$, vector of means during periods of normal interest rates ($\boldsymbol{\mu}_1$) and in a low interest rate environment ($\boldsymbol{\mu}_2$), and decay parameter $\lambda$. Panel (B) presents estimates of the covariance matrix $\boldsymbol{\Sigma}_{\eta}$, and the transition probabilities $p_{11}$ and $p_{22}$. Bold entries denote parameter estimates significant at the 5 percent level. Standard errors appear in parentheses.

\end{table}

\end{document}