Fixed effects regression is doing something I'm not noticing?

In Stata I complete a fixed-effects conditional logistic regression model of a binary predictor (1==Unemployed | 0 ==Employed) on a binary outcome (1==Overweight | 0 == Not overweight), with some controls in a longitudinal panel of 3 waves.

Code:

  clogit kidsweight i.parentsunemployed i.urban_or_rural i.year i.parents_age_y i.Parents_Educa i.Parents_Marital, cluster (id) group(id) nolog
note: multiple positive outcomes within groups encountered.
note: 9,091 groups (23,274 obs) dropped because of all positive or
      all negative outcomes.

Conditional (fixed-effects) logistic regression

                                                Number of obs     =      5,532
                                                Wald chi2(12)     =     268.06
                                                Prob > chi2       =     0.0000
Log pseudolikelihood = -1892.4384               Pseudo R2         =     0.0603

                                                       (Std. Err. adjusted for 1,945 clusters in id)
----------------------------------------------------------------------------------------------------
                                   |               Robust
       kidsweight|      Coef.   Std. Err.      z    P>|z|     [95% Conf. Interval]
-----------------------------------+----------------------------------------------------------------
               1.parentsunemployed |   .2586795   .0991969     2.61   0.009     .0642571    .4531019
                      1.urban_or_rural |   .0284788   .1521921     0.19   0.852    -.2698122    .3267699
                                   |
                              year |
                                1  |    .331549   .0608113     5.45   0.000     .2123611    .4507368
                                2  |  -.5641933   .0786183    -7.18   0.000    -.7182823   -.4101043
                                   |
                       parents_age_y |
                            30-39  |   -.019373   .1321831    -0.15   0.883    -.2784471     .239701
                       40 or more  |   -.131816   .1917598    -0.69   0.492    -.5076582    .2440262
                                   |
               Parents_Educa |
Leaving Certificate to Non Degree  |   .3654921   .2249296     1.62   0.104    -.0753619     .806346
        Primary Degree or greater  |   .4395884   .2934593     1.50   0.134    -.1355812    1.014758
                                   |
                     Parents_Marital |
                                2  |   -.154054   .2966866    -0.52   0.604     -.735549    .4274409
                                3  |  -.4093562   .3844533    -1.06   0.287    -1.162871    .3441584
                                4  |  -.1921434   .1805024    -1.06   0.287    -.5459217    .1616349
                                5  |   .7150017   1.125252     0.64   0.525    -1.490451    2.920455
----------------------------------------------------------------------------------------------------

. margins, dydx(parentsunemployed) post

Average marginal effects                        Number of obs     =      5,532
Model VCE    : Robust

Expression   : Pr(kidsweight |fixed effect is 0), predict(pu0)
dy/dx w.r.t. : 1.parentsunemployed

-----------------------------------------------------------------------------------------------
                              |            Delta-method
                              |      dy/dx   Std. Err.      z    P>|z|     [95% Conf. Interval]
------------------------------+----------------------------------------------------------------
            1.parentsunemployed|   .0605013   .0229353     2.64   0.008     .0155489    .1054537
-----------------------------------------------------------------------------------------------
Note: dy/dx for factor levels is the discrete change from the base level.

I report the coefficient on the unemployment variable as the effect of unemployment on overweight. So, here I say if your parent experienced unemployment at any point across the three waves of the study, your probability of being overweight was 0.06 percentage points higher.

I received a comment that I treat transitions from unemployment to employment on weight similarly to transitions from employment to unemployment on weight.

But, I only ever report the coefficient on parentsunemployed (0.06) and it's my understanding that due to how I set up my binary predictor and outcome I'm only ever considering the effect of a change from employment to unemployment on a change from not overweight to overweight.

So, why were changes from unemployment to employment even mentioned? Is it possible that I am considering this and don't even know it? And how?!

I could really do with some advice!

All the best,

John