It depends whether your data from various waves of the survey refer to the same units of observation (households, other entities...) or these units were changed in each wave. I guess, the former is rather implausable looking at the time distance between your observations. In this case you could use usual regression approach, but make sure to deflate the monetary values from different time periods to comparable units (use real values). However, with this approach you assume that the relationships involved are constant over time. With growing time span of the analysis this assumption weakens. On the other hand, if the observations from different years refer to the same units you could use panel data regression.
There is no problem with having data every 5 years; this changes the unit of time but otherwise is not a problem. If you used annual data, then you would be using summary data as compared with daily data. Since the time intervals are uneven, you should include a measure of time (it could be the actual year, or it could start with 0 in your first year and then be based on the number of years since the first year). Blazej Lyszczarz is correct that you should include an inflation adjustment for monetary values. If you have only 10 observations, that is a significant limitation, and you need to be wary of overfitting the data.
It depends what you want to analyse. If you want to explain relations between variables of the same dimension and you use also ratios as explaining variables no transformation to real varables is necessary. For absolute variables, instead of disinflating them, one could include a price index as explaining variable. In any case, before regressing, one should have a close look at the data (preferably by diagrams). I am skeptical, that regressions will bring valuable results with only 10 data points over quite a long time span, because the assumption of constant relations (no breaks) seems too strong in this case.