How to interpret a dummy variable for "year"?

[Kaine Dinh]

Hello,
I am running a regression model as below:

lprice = 6.3 - 0.01lpassen - 0.88ldist + 0.1ldistsq + 0.02y2018 + 0.009y2019 + 0.07y2020
Where:
lprice = the average flight ticket in logarithm
lpassen = average number of passengers per day in logarithm
ldist = distance in logarithm
ldistsq = ldist^2
y2018, y2019, y2020 = dummy variables for year 2018, 2019 and 2020, respectively.

I would like to ask that how can I interpret the coefficient of the variables y2018, y2019 and y2020? Because if I interpret like normal: a change in the year 2018 by one unit is associated with a 2% change in the average ticket price would sound quite confusing to me.
Thank you for your help in advance!

[Clyde Schechter]

I assume that by a "dummy" variables for years 2018, 2019, or 2020, respectively, you mean, variables that take on the values 1 in observations that refer to things happening in year 2018 (resp. 2019, 2020) and 0 in observations referring to other years. In addition there must be some other reference year for which there is no corresponding variable, and for observations referring to that year, all three of the variables y2018, y2019, and y2020 are zero. That is the usual (but not only) way of creating indicator variables.

So a difference of 1 in y2018 means a difference between the year being 2018 and the year being the reference year. So let's say your reference year is 2021. (The most common choice for a reference year is either the first or last year that occurs in the data, but any year in the data can be used for the purpose.) Then your model is saying that lprice is 0.02 higher in year 2018 than it is in year 2021, all else being equal. (And since lprice is the log of price, that corresponds approximately to price being 2% higher in year 2018 than in year 2021.) Similar interpretations apply to y2019 and y2020.

[Kaine Dinh]

Oh thank you for your clarification! It helps me a lot to understand the problem. Best wishes!