Fixed effects model/interpretation

[William Vallelonge]

Hi all,

I am doing a fixed effects model in stata in order to find out what factors affect home prices. I am in doubt whether I type in the right command to stata AND how to interpret the results so I hope some of you can help me.

WHAT I TYPE IN:
1) Make stata understand i use quarters:

gen qdate = quarterly(TIME, "YQ")
format qdate %tq

2) Running the fixed effect regression

xtset ID qdate
xtreg Homeprices wage unemploymentrate interest buildingcosts population qdate, fe

WHAT I GET AS RESULT:

With respect to the result, is it then right to say that: As unemployment rate increases by 1 % (unemployment rate is measured in percentages), home prices (measured in an index) decrease by 108 %. Of course this does not sound right, but what am I doing wrong?

Any help is very much appreciated. Thank you very much in advance.

Best regards
William

[FernandoRios]

Hi WIlliam,
it would be useful for you to have a table with the summary statistics of your variables, so you can better interpret your results. For example, you say that unemployment is measured in percentage. Does it mean that if unemployment rate is 5% it is stored as 5? or is it stored as 0.05?
Regarding the impact on Home prices, are home prices in logs or levels? If its in levels, the impact would be a reduction of 108$ not %. Only if it is in logs could you interpret it as a percentage change effect.
Finally, it is useful to also estimate the model without Fixed effects, to have an easier model to interpret and see if the results are still consistent with each other.
Best Regards
Fernando

[William Lisowski]

I think it is more correct to say "a change of +1 in the value of Unemploymentrate will produce a decrease -108 in the home price index."

Since you do not tell us how the home price index is measured - what range of values does it take? Is it an "index" in the sense of being an average home price measured in the local currency? Is it measured in the local currency adjusted for inflation? - I cannot say whether the home price index decreasing by 108 is sensible or not.

[William Lisowski]

No, you cannot say "when unemployment increases by 1 percent, home prices decrease by 1,08%".

You can say "when the unemployment rate increases by 1 percentage point, the home price index decreases by 1,08". For example, the index would decrease from 121.00 to 119.92, which is a decrease of just 0.89% of 121. Or it would decrease from 81 to 79.92, which is a decrease of 1.33% of 81.

I am distinguishing between "unemployment" and "the unemployment rate" - the former is the number of unemployed persons in the labor force, the latter is the percentage of the labor force that is unemployed.

I am distinguishing between a 1.08 percent change in home prices (as measured by the home price index) and a change of 1.08 in the home price index.

Your model as you have constructed it does not predict percentage change.

Your commands are syntactically correct for Stata. To judge whether they are appropriate for the objectives of your analysis would require much more understanding of your data and your objectives.