Combining stacked barchart (with negative values) with line using twoway

[Andy Prenner]

Dear all,

I'm trying to overlay a stacked bar chart with a line chart.

I plot 7 variables (some of which are negative most of the time) in the barchart with the "graph bar, stack" routine:

graph bar (asis) var1 var2 var3 var4 var5 var6 var7, stack

I now want to add a variable that is the difference between all positive and negative variables as a line to the chart.

It seems that "graph bar" does not allow this and that I'd need to switch to a "twoway" routine with "bar", "rbar" and "line".

However, trying this out and reading the discussion HERE, I'm not sure how this could be done if some of my variables are negative - or if this can be done in my situation at all.

[Stephanie Stewart]

I am running into the same problem as Andy, so I hope that someone might have a solution for how to create a bar chart with both positive and negative values.

This is what the graph looks like using graph bar but obviously this won't allow me to plot another line. You can see that 3 of the four components oscillate between being positive and negative, so I can't combine them in any one specific order to ensure there won't be any overlap. I also am trying to use the over() option with a quarterly date variable, but I can't figure out how to only display some of the labels. If I have to stick with bar graph, then it would be helpful to have an idea of how to do this.

This is the code I currently have, but obviously the way I am generating the bars for the twoway plot is not correct given that not all the values are positive.

graph bar (asis) comp_ugap comp_pc_urate comp_infl_exp comp_pc_infl if inrange(date_yq, tq(1984q1), tq(2019q4)), over(date_yq, label(angle(45) alternate labsize(tiny))) bargap(.1) name(inflation_components_v1, replace) stack title("Inflation Decomposition")  legend(order(1 "Unemployment Gap" 2 "PC of Unemployment Gap" 3 "Inflation Expectations" 4 "PC of Inflation Deviations"))

* Generate bars for the decomposition chart
gen bar_pc_infl = comp_pc_infl
gen bar_infl_exp = comp_infl_exp + comp_pc_infl 
gen bar_pc_urate = comp_pc_urate + comp_infl_exp + comp_pc_infl
gen bar_ugap = comp_ugap + comp_pc_urate + comp_infl_exp + comp_pc_infl
       
twoway (bar bar_ugap date_yq, barwidth(.75)) || ///
       (bar bar_pc_urate date_yq, barwidth(.75)) || ///
       (bar bar_infl_exp date_yq, barwidth(.75)) || ///
       (bar bar_pc_infl date_yq, barwidth(.75)) || ///
       (line cpi_infl_yoyUSA date_yq) if inrange(date_yq, tq(1984q1), tq(2019q4)), title("Inflation Decomposition") legend(order(1 "Unemployment Gap" 2 "PC of Unemployment Gap" 3 "Inflation Expectations" 4 "PC of Inflation Deviations" 5 "YoY CPI Inflation")) xlabel(,angle(45))

Here is a snippet of the data from 2010-2019:

clear
input double date_yq float(cpi_infl_yoyUSA infl_exp ugap pc_infl pc_urate b_pc_infl b_infl_exp b_pc_urate b_ugap comp_pc_infl comp_infl_exp comp_pc_urate comp_ugap)
201   1.767765 1.8775   4.789333  1.1343961   .8520799 .3012544 .850033 -.12143505 -.072333366   .3417418  1.595937  -.10347237   -.3464286
202  1.1756089 1.7209  4.6296663  -.2795476  .17345235 .3012544 .850033 -.12143505 -.072333366 -.08421494  1.462822 -.021063196   -.3348793
203   1.270248  1.609      4.671  -.5092422  2.2899835 .3012544 .850033 -.12143505 -.072333366 -.15341145  1.367703  -.27808428   -.3378691
204   2.141127 1.7247  4.2133336   .8880997 -1.9988465 .3012544 .850033 -.12143505 -.072333366   .2675439 1.4660518   .24273004   -.3047646
205  3.4303954 2.1304  4.2556667  1.4024277 -4.3533335 .3012544 .850033 -.12143505 -.072333366   .4224875 1.8109102    .5286473   -.3078267
206   3.756174 2.0125      4.198    .826742  -.8935137 .3012544 .850033 -.12143505 -.072333366  .24905966 1.7106915   .10850388   -.3036555
207  3.2937765 1.9692  3.8413334  1.2592603   1.966702 .3012544 .850033 -.12143505 -.072333366   .3793577  1.673885  -.23882657  -.27785656
208   2.815192  2.075   3.484666   .3722294   .6017069 .3012544 .850033 -.12143505 -.072333366  .11213575 1.7638185  -.07306831  -.25205764
209   1.889765 2.1831      3.429 -1.1387343  .36911225 .3012544 .850033 -.12143505 -.072333366  -.3430487  1.855707  -.04482317   -.2480311
210  1.6977844 2.1098   3.271334   .3159398 -1.0183913 .3012544 .850033 -.12143505 -.072333366  .09517825 1.7933997    .1236684   -.2366266
211  1.8893654 2.1931      3.048   .7706487   .6093867 .3012544 .850033 -.12143505 -.072333366   .2321613 1.8642073  -.07400091  -.22047213
212   1.681829    2.1   2.990333  -.9458265   2.596865 .3012544 .850033 -.12143505 -.072333366  -.2849344 1.7850692   -.3153504  -.21630086
213  1.3928498 2.0414   2.799333  -.8478221   2.510469 .3012544 .850033 -.12143505 -.072333366 -.25541013 1.7352573   -.3048589   -.2024852
214  1.5533587 1.8626   2.507333   .5671202  .09379639 .3012544 .850033 -.12143505 -.072333366  .17084745 1.5832714  -.01139017  -.18136385
215   1.233471  1.984  2.2153335  -.5762158  -.6486745 .3012544 .850033 -.12143505 -.072333366 -.17358755 1.6864654   .07877183  -.16024253
216  1.4054564 1.8881  1.9556665 -.28374568  1.1163388 .3012544 .850033 -.12143505 -.072333366 -.08547963 1.6049473  -.13556267  -.14145994
217  2.0508456 1.9583  1.4959998  1.0971627  -.9334263 .3012544 .850033 -.12143505 -.072333366   .3305251 1.6646196   .11335067   -.1082107
218   1.783154  2.099  1.3696666   .5602407 -.50588816 .3012544 .850033 -.12143505 -.072333366  .16877496 1.7842193   .06143256   -.0990726
219  1.2480277 1.9162  1.0099998  -.4613005  .23187394 .3012544 .850033 -.12143505 -.072333366  -.1389688 1.6288333 -.028157625  -.07305668
220 -.06269593 1.8938   .8493333 -1.9720105    .896744 .3012544 .850033 -.12143505 -.072333366  -.5940768 1.6097925  -.10889616  -.06143514
221 -.03827201 1.9731   .7563334  -.0876354   .1153743 .3012544 .850033 -.12143505 -.072333366 -.02640055    1.6772 -.014010483  -.05470814
222   .1095034 2.0422   .4289999  .14783333 -1.2434194 .3012544 .850033 -.12143505 -.072333366  .04453544 1.7359375    .1509947 -.031031007
223   .4662646 2.0435   .3693333  .58983105  .08079761 .3012544 .850033 -.12143505 -.072333366   .1776892 1.7370423 -.009811662  -.02671512
224  1.0802671 1.9849  .24300003  .19646227  1.0364263 .3012544 .850033 -.12143505 -.072333366  .05918512 1.6872305  -.12585849  -.01757701
225  1.0470631 2.0574   .2833333  -.7048873    .196718 .3012544 .850033 -.12143505 -.072333366  -.2123504  1.748858  -.02388846  -.02049445
226  1.1176153 2.2276  .25699997  -1.342066  -.5964288 .3012544 .850033 -.12143505 -.072333366  -.4043033 1.8935336   .07242736 -.018589674
227  1.8006208 2.2029  .13066673  -.7500671   .7978568 .3012544 .850033 -.12143505 -.072333366   -.225961 1.8725376  -.09688778 -.009451564
228  2.5393205 2.3596 -.06133318  2.4358275   .5196755 .3012544 .850033 -.12143505 -.072333366   .7338037 2.0057378  -.06310682  .004436435
229  1.9019914  2.275 -.25433302  .16263594  -.8419599 .3012544 .850033 -.12143505 -.072333366  .04899479  1.933825   .10224345  .018396763
230   1.966925 2.2053 -.27966642 .069986925   1.259212 .3012544 .850033 -.12143505 -.072333366  .02108387  1.874578  -.15291247  .020229213
231  2.1175575 2.0907  -.4383335    .354585  -.5408554 .3012544 .850033 -.12143505 -.072333366  .10682028  1.777164  .065678805   .03170614
232  2.2141945 2.2117  -.5636668  -1.298792 -1.5147632 .3012544 .850033 -.12143505 -.072333366  -.3912668  1.880018   .18394536   .04077192
233   2.711887 2.2597  -.6546664 -.57885927  -.8058838 .3012544 .850033 -.12143505 -.072333366  -.1743839 1.9208196   .09786254   .04735423
234    2.64094 2.2589  -.8123333  1.4723552  -.9567183 .3012544 .850033 -.12143505 -.072333366   .4435535 1.9201394   .11617914   .05875881
235  2.2031314 2.3658  -.7356665   .8729657   .7209505 .3012544 .850033 -.12143505 -.072333366  .26298475  2.011008  -.08754867   .05321323
236   1.644936 2.2651  -.6923335  -.3188704 -.04583866 .3012544 .850033 -.12143505 -.072333366  -.0960611 1.9254097   .00556642   .05007881
237  1.8113756 2.0855  -.9146664   .7916945  -1.063224 .3012544 .850033 -.12143505 -.072333366  .23850144 1.7727438   .12911266    .0661609
238  1.7574885 2.0369  -.9370003  -.9528717 -1.0080968 .3012544 .850033 -.12143505 -.072333366  -.2870568 1.7314322    .1224183   .06777638
239   2.032914 2.1384  -.9250002 -1.0397493   -.311735 .3012544 .850033 -.12143505 -.072333366   -.313229 1.8177106   .03785555  .066908374
240          .      .          .          .          . .3012544 .850033 -.12143505 -.072333366          .         .           .           .
241          .      .          .          .          . .3012544 .850033 -.12143505 -.072333366          .         .           .           .
242          .      .          .          .          . .3012544 .850033 -.12143505 -.072333366          .         .           .           .
end
format %tq date_yq

[Nick Cox]

I gather that these components are considered to be additive, but not only they often negative, they seem to have quite different magnitudes.

Here is a different take. Anyone liking the main idea would probably want to move on to different colours.

rename (comp_ugap comp_pc_urate comp_infl_exp comp_pc_infl) (comp#), addnumber

reshape long comp, i(date_yq) j(which)

label def which 1 "Unemployment Gap" 2  "PC of Unemployment Gap" 3  "Inflation Expectations" 4  "PC of Inflation Deviations"

label val which which

twoway bar comp date_yq, by(which, yrescale col(1) note("")) ytitle(%) xla(200(4)240, format(%tq)) xtitle("") fcolor(stc1*0.2)

[Stephanie Stewart]

Shortly after posting I found this post on Statalist: https://www.statalist.org/forums/for...nd-a-line-plot

I was able to get the chart that I needed with the following adaptations:

gen top = 0 
gen bottom = 0 

foreach j in infl_exp pc_urate ugap pc_infl { 
    
    gen base_`j' = cond(comp_`j' >=  0, top, bottom)
    gen show_`j' = base_`j' + comp_`j'
    
    replace top = top + comp_`j' if comp_`j' >= 0 
    replace bottom = bottom + comp_`j' if comp_`j' < 0 
}

local opts barw(0.75)
twoway (rbar base_ugap show_ugap date_yq, `opts') || /// 
(rbar base_pc_urate show_pc_urate date_yq, `opts') || ///
(rbar base_infl_exp show_infl_exp date_yq, `opts') || ///
(rbar base_pc_infl show_pc_infl date_yq, `opts') || ///
(line cpi_infl_yoyUSA date_yq) if inrange(date_yq, tq(1984q1), tq(2019q4)) ,  title("Inflation Decomposition") legend(order(1 "PC of Inflation Deviations" 2 "PC of Unemployment Gap" 3 "Inflation Expectations" 4 "Unemployment Gap" 5 "YoY CPI Inflation")) xlabel(,angle(45))

[Stephanie Stewart]

Thanks for the suggestion, Nick. Part of the point of the graph is to highlight the different magnitudes that you noted. Your method from a different post was helpful in getting what I needed, though.