There are two ways to measure the value of money in the past.
1) As Victor mentioned there are inflation statistics covering the last 100 or so years that value the currency against an ever-changing basket of goods. This is sufficient when measuring general inflation over the period of the hundred years where there is data. This is how it was measured in your example.
2) For older time periods or where a value comparison is required between specific items (particularly where these were not in the basket of goods used for the inflation calculation) Historical records of the price of comparable goods can be used. This is in effect the same as mark to market valuations for illiquid financial instruments and requires poring through records to find the price of either a comparable basket of goods to one that would be used for inflation calculations today or a comparable set of items. An example of this is finding the value of a particular type of house (say a terraced house in London) in the 19th Century compared to the same house today by finding records of how much comparable houses would sell for, on average, then and now.
This second measure is also used where the country in question didn't or doesn't keep reliable inflation statistics which may well be true of Colombia in the 90s. This means that there is a chance that this way of estimating Escobar's wealth in today's terms may have also been used.
Another notable reason to use this methodology is that (unless you are using exchange rates in purchasing power parity terms) the value of money held in different currencies is different. This is even true today as the value of $1 in INR in India is likely to be higher than the value of a dollar in the US in terms of what you can buy. Using this methodology allows for a more accurate comparison in values where different countries and currencies are involved.