M. Kennet, "A Structural Model of Aircraft Engine Maintenance", Journal of Applied Econometrics, 1994, Vol. 9, No. 4, pp. 351-368. The datafiles JT8REG.ASC, JT8DREG.ASC, JT9REG.ASC, and JT9DREG.ASC are the final, "clean" ASCII data used in this paper. The filenames are defined as follows: JT8REG.ASC = data from JT8D engines, regulated years (1964-1977) JT8DREG.ASC = data from JT8D engines, unregulated years (1977-1988) JT9REG.ASC = data from JT9D engines, regulated years JT9DREG.ASC = data from JT9D engines, unregulated years. The data in the files are unformatted. Each row contains one observation as follows: CHOICE STATE DSTATE AL2 AL3 AL4 AL5 AL6 AL7 where CHOICE = trichotomous variable taking values 0 do nothing 1 convenience engine removal 2 shop visit engine removal STATE = discretized state variable taking values 1 through 88, with odd values indicating no engine shutdown currently in history and even values indicating that the shutdown has occurred. For example, State 1 indicates that the engine has less than 795 hours of flight time with no engine shutdown; State 2 indicates less than 795 hours of flight time, but with an engine shutdown. The value of the state variable reverts to unity immediately following a shop visit. DSTATE = difference between this month's state variable and last month's. AL2 through AL7 = indicators for Airlines numbered 2 through 7. Note that there are only four airlines (i.e., AL2 through AL4) in the sample for JT9D engines. Each observation represents one engine in one calendar month. A complete shop visit cycle begins in States 1, 2, 3, or 4 with Choice 0 and continues to the point where we observe a Choice 2, at which point the cycle is regenerated. See Kennet (1994) for further details, or contact the author at kennetm@ore.psb.bls.gov Mark Kennet Economist Office of Economic Research U.S. Bureau of Labor Statistics Suite 4915 2 Massachusetts Avenue, N.E. Washington, DC 20212