Obviously I'm talking about a theoretical graph not one where you can just plot the data. The only two options I can think of are just using shapes in word/powerpoint or using TikZ in Latex which is very time consuming.

I would be interested to know what your preferred methods are for drawing good looking digital graphs.

  • 2
    $\begingroup$ Ipe is great. Allows you to add latex code and export to pdf (and edit later on). $\endgroup$
    – tdm
    May 11 at 10:00
  • $\begingroup$ Why not simply generate the datapoints and then plot them in your favourite data visualization program? $\endgroup$ May 11 at 10:06
  • 1
    $\begingroup$ @HenryM I have read the question. I am just not seeing what kind of theoretical graph there would exist that you cannot create by simply making up the data and then plotting that instead. AFAIK any graph is just a representation of (possibly transformed) numbers. $\endgroup$ May 11 at 10:15
  • 1
    $\begingroup$ @Henry Why would you assume that MaartenPunt's 'favourite data visualization program' is 'excel'? Personally I do what MaartenPunt suggest and simply code up the functions make the points for the graph and plot it, this I prefer to do in R. Not saying it is ideal, but having learned the software the trade-off between prettyness and speed of production usually favors just sticking to my 'favourite data visualization program'. $\endgroup$ May 11 at 11:00
  • 1
    $\begingroup$ Ipe is what I used for formal purposes, but I also like Desmos. Here is an example. There is definitely a learning curve, and the main focus of the site is interactivity, it is not an all purpose programming language. $\endgroup$
    – Giskard
    May 11 at 14:44

Give @Giskards comments I'll put up my answer in the comments as answer. I usually fake the data and then use my favourite plotting program. For example here is the requested income substitution graph challenge created with R and GGplot2. For completeness I add the code, but given that it's quite long I add it to the end of my answer. Added bonus of GGplot/R is that if one is well versed with it it is simple to include LaTEX annotation.

enter image description here

Making a similar graph in Excel is also doable. I'll admit though that Excel becomes a pain once we want multiple axes and scales, and the same probably applies to GGPlot. And of course Excel doesn't work well with LaTEX.

Code to generate graph (all parameters can be adjusted, the only thing that needs to be fixed afterwards is the vertical position of the arrows and the text).

#Using a CD utility function for convenience and because it generates easy demands
#As the A in that function is just for scaling we'll set it to 1
#Exponent on utility function
alpha <- 0.5
#Price of goods
p_x <-1
p_y <-1
#Price of x after price increase
p_x2 <-4
M <- 10

#Basic numbers as input

#Budget constraint 1 & 2 (formulated in y-x plane)
df$BC_1 <-(M-p_x*df$x_amount)/p_y
df$BC_2 <-(M-p_x2*df$x_amount)/p_y

#Calculating utility for correct indifference curves
#By the marginal spending rule alpha/(1-alpha)*(y/x)=p_x/p_y
#Therefore y=((1-alpha)/alpha)*(p_x/p_y)*x
#By the budget constraint y=(M-p_x*x)/p_y
#Using that and solving gives x=(M*alpha)/p_x and y=((1-alpha)*M)/p_y
#Utility indifference curve 1
#Utility indifference curve 2

#Indifference curve 1
#Indifference curve 2

#To find the final budget constraint first identify the x and y value
#To do this we set the first derivative of indifference curve 1 equal to 
#the new -p_x2/p_y and find its root 
f1 <- function (x) -alpha*(U1/x^alpha)^(1/(1-alpha))/((1-alpha)*x)+p_x2/p_y

Solution_x <- uniroot(f1,c(0.1,20))$root
Solution_y <- (U1/(Solution_x^alpha))^(1/(1-alpha))
#Determine the intercept

f2 <- function(M) M-p_x2*Solution_x - p_y * Solution_y
M2 <- uniroot(f2, c(0,30))$root

df$BC_3 <- M2/p_y - p_x2/p_y * df$x_amount

#Reshaped dataframe to long format
df_2 <- data.frame(x_amount=rep(df$x_amount, times=5),
                   y_variables=c(df$BC_1,df$BC_2,df$indif1,df$indif2, df$BC_3),
                   role=factor(rep( colnames(df)[2:6], each=length(df$x_amount)))

ggplot(data = df_2) +
  aes(x_amount, y_variables,colour=role)+
  geom_vline(xintercept = Solution_x, size=1, linetype=3)+
  geom_vline(xintercept = (M*alpha)/p_x , size=1, linetype=3)+
  geom_vline(xintercept = (M*alpha)/p_x2, size=1, linetype=3)+
  geom_segment(x=(M*alpha)/p_x, y=3,yend=3, xend=Solution_x, colour="black", arrow = arrow(length = unit(0.25, "cm")))+
  geom_segment(x=Solution_x, y=5,yend=5, xend=(M*alpha)/p_x2, colour="black" ,arrow = arrow(length = unit(0.25, "cm")))+
  geom_segment(x=(M*alpha)/p_x, y=0.8,yend=0.8, xend=(M*alpha)/p_x2, colour="black", arrow = arrow(length = unit(0.25, "cm")))+
  annotate("text",x=3.2, y=3.7, label = "Substitution effect", size=3)+
  annotate("text",x=1.9, y=5.3, label = "Income effect", size=3)+
  annotate("text",x=3, y=1.2, label = "Total effect", size=3)+
  ylab("Amount of y")+
  xlab("Amount of x")+
  theme(legend.position = "none", 
        panel.background = element_rect(fill = NA), 
        panel.border = element_rect(linetype = 1 , fill = NA))

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.