The APCD model = Age Period Cohort – Detrended

Louis Chauvel

 

Site : www.louischauvel.org/apcdex

email : chauvel@louischauvel.org

 

 

This web page intends to present the STATA “ssc install apcd” ado file, its methodological background and some examples. 

 

1. Problems:

 

The problem with APC models is the general lack of stability of linear trends of the age, period and cohort estimates where appropriate constraints should be added in order to obtain stable estimates, but there are still debates on this choice of coefficients. In their APC_IE methodology, Yang and al. propose that a Principal Component Analysis solve this problem, but the usefulness of this choice is still debated (O’Brien 2011).  The “detrended cohort estimates” DCE estimates (see 2. Methods) proposes a separation between linear trend analysis (that definitively can not be disentangled between age, period and cohort linear effects) and fluctuations. This model is able to tackle better the trends problem that other models generate, for example with education where the APC-IE proposes a steady decline of educational level with age; in reality, the linear trends of APC coefficients are generally not interpretable (see part 5 below).

 

2. Methods:

Please download the pdf: www.louischauvel.org/apcdmethodo.pdf where the main concepts are presented with the model itself. 

 

 

 

3. First example: a cohort based variable: veterans

 

See the do file:

www.louischauvel.org/apcdvet.do

[here is an extract of the Ipums US census and ACS 1980-2010, Steven Ruggles, J. Trent Alexander, Katie Genadek, Ronald Goeken, Matthew B. Schroeder, and Matthew Sobek. Integrated Public Use Microdata Series: Version 5.0 [Machine-readable database]. Minneapolis: University of Minnesota, 2010.]

The status of veteran is deeply connected to birth cohorts. About 45% of the male population born in 1947 was drafted for the Vietnam War while less than 10% of those born before 1940 or after 1955 were. Wars are deep contexts of cohort effect formation and it is why APC analysis of veterans make sense, but more as a test of a cohort method than as a plan to make new discoveries.

The first step here is to test the apcd model on Vietnam war veterans (viet=0/1) in a logit APC-D model.

 

The contextual variables give significant elements, first for gender (!) then for ethnicities: drop outs, Hispanics, other races (that are social groups specific of a more or less recent entry in the US territory) show lower rates of veterans. We notice an inverted U-curve of veteran distribution by educational level (see Card, 200x on this point). 

 

 

*******************************************

#######       APC Detrended model   #######

*******************************************

 

(623178 missing values generated)

 

Iteration 0:   log pseudolikelihood = -1.303e+08 

Iteration 1:   log pseudolikelihood = -1.060e+08 

Iteration 2:   log pseudolikelihood = -1.051e+08 

Iteration 3:   log pseudolikelihood = -1.051e+08 

Iteration 4:   log pseudolikelihood = -1.051e+08 

 

Generalized linear models                          No. of obs      =    571855

Optimization     : ML                              Residual df     =    571833

                                                   Scale parameter =         1

Deviance         =  210179661.9                    (1/df) Deviance =  367.5543

Pearson          =   1071985128                    (1/df) Pearson  =  1874.647

 

Variance function: V(u) = u*(1-u)                  [Bernoulli]

Link function    : g(u) = ln(u/(1-u))              [Logit]

 

                                                   AIC             =  367.5402

Log pseudolikelihood = -105089830.9                BIC             =  2.03e+08

 

 ( 1)  [viet]coh_1920 + [viet]coh_1930 + [viet]coh_1940 + [viet]coh_1950 + [viet]coh_1960 + [viet]coh_1970 = 0

 ( 2)  - 5*[viet]coh_1920 - 3*[viet]coh_1930 - [viet]coh_1940 + [viet]coh_1950 + 3*[viet]coh_1960 + 5*[viet]coh_1970 = 0

 ( 3)  [viet]age_0030 + [viet]age_0040 + [viet]age_0050 + [viet]age_0060 + [viet]age_0070 = 0

 ( 4)  - 4*[viet]age_0030 - 2*[viet]age_0040 + 2*[viet]age_0060 + 4*[viet]age_0070 = 0

 ( 5)  [viet]per_1980 + [viet]per_1990 + [viet]per_2000 + [viet]per_2010 = 0

 ( 6)  - 3*[viet]per_1980 - [viet]per_1990 + [viet]per_2000 + 3*[viet]per_2010 = 0

------------------------------------------------------------------------------

             |               Robust

        viet |      Coef.   Std. Err.      z    P>|z|     [95% Conf. Interval]

-------------+----------------------------------------------------------------

    coh_1920 |  -.9707834   .0232047   -41.84   0.000    -1.016264    -.925303

    coh_1930 |  -.3195585   .0235933   -13.54   0.000    -.3658006   -.2733165

    coh_1940 |   1.115685    .015823    70.51   0.000     1.084672    1.146697

    coh_1950 |   1.658306   .0150098   110.48   0.000     1.628887    1.687725

    coh_1960 |  -.5315142   .0242539   -21.91   0.000    -.5790511   -.4839773

    coh_1970 |  -.9521343   .0243227   -39.15   0.000    -.9998058   -.9044627

    age_0030 |   .0117883   .0099741     1.18   0.237    -.0077605    .0313371

    age_0040 |  -.0218496   .0118446    -1.84   0.065    -.0450646    .0013655

    age_0050 |    .024114   .0131639     1.83   0.067    -.0016867    .0499147

    age_0060 |  -.0298326   .0131493    -2.27   0.023    -.0556049   -.0040604

    age_0070 |   .0157799   .0109327     1.44   0.149    -.0056478    .0372076

    per_1980 |   .0050198   .0076948     0.65   0.514    -.0100618    .0201014

    per_1990 |  -.0105417   .0112558    -0.94   0.349    -.0326027    .0115193

    per_2000 |   .0060241   .0115525     0.52   0.602    -.0166184    .0286666

    per_2010 |  -.0005021     .00784    -0.06   0.949    -.0158683     .014864

    rescacoh |  -.6195064   .0392983   -15.76   0.000    -.6965297    -.542483

    rescaage |   .0197929   .0126496     1.56   0.118    -.0049998    .0445856

      hispan |  -.3166048   .0304606   -10.39   0.000    -.3763066   -.2569031

          aa |  -.0960236   .0226548    -4.24   0.000    -.1404262    -.051621

         sex |  -2.464644   .0198651  -124.07   0.000    -2.503579   -2.425709

       orace |  -.4056752   .0317338   -12.78   0.000    -.4678723    -.343478

    _Ieduc_5 |   .4058921    .048345     8.40   0.000     .3111377    .5006465

    _Ieduc_6 |   .8972972   .0272275    32.96   0.000     .8439322    .9506621

    _Ieduc_7 |   1.130047   .0298255    37.89   0.000      1.07159    1.188504

    _Ieduc_8 |   1.200131   .0318622    37.67   0.000     1.137682     1.26258

    _Ieduc_9 |    .722688    .030655    23.57   0.000     .6626053    .7827706

   _Ieduc_10 |   .6184479   .0318983    19.39   0.000     .5559285    .6809674

       _cons |  -.8965447   .0319209   -28.09   0.000    -.9591085   -.8339808

------------------------------------------------------------------------------

(36 missing values generated)

(36 missing values generated)

 

 

*******************************************

Delta Bic = -27506167

*******************************************

 

 

 

 

Figure3.1: shape of cohort coefficients of Vietnam veterans 

 

 

 

4. second example: education

 

Education is, in the set of common variables, one of the most influenced by birth cohort. The period of entry in the age of choice between following education or finding one’s independent life is strategic, in terms of opportunities or limits.

www.louischauvel.org/apcdcpseduc.do  

[here is an extract of 1975-2010 (each 5 years) March CPS extracts source IPUMS, see: Miriam King, Steven Ruggles, J. Trent Alexander, Sarah Flood, Katie Genadek, Matthew B. Schroeder, Brandon Trampe, and Rebecca Vick. Integrated Public Use Microdata Series, Current Population Survey: Version 3.0. [Machine-readable database]. Minneapolis: University of Minnesota, 2010.]

 

 

Figure4.1: % of BA degree owners (or higher) at age 45 by birth cohort

 

 

These data show the extreme acceleration in educational resources of the cohorts born in early baby boom (circa 1950). David Card (xxxx) and Robert Mare (xxxx) have documented this singularity; the first one insist in the Vietnam war context that increased the incentive to follow education, and the second one on the composition effects of immigration.

 

When we control by usual contextual variables, the APCD model shows the specificity of the early baby boom. These specific traits has been largely commented (xxx)

 

Figure4.2: shape of cohort coefficients of BA degree owners (or higher) in the USA

 

 

 

The intensity of the cohort fluctuations for MA owners is even stronger, and it is amazing that these gaps have received so modest interest in the sociological literature. For MA owners, a part of the gap is absorbed over life course (H=-.16), that is significant. This trend of slight hysteresis could be due to the fact that cohorts with lower achievements in terms of Ma degree can (partially) catch up later.

 

 

Figure4.3: % of MA degree owners (or higher) at age 30, 40 & 50 by birth cohort

Figure8: shape of cohort coefficients of MA degree owners (or higher) in the US

 

 

This means that birth cohort is an important parameter of inequality of distribution of education, since the context of educational development between age 17 and 23 is of major importance for individual’s opportunities. More precisely, the cohorts born in the late 1940’s, beginning of the 1950’s, have benefited from exceptional opportunities for having longer education. A major aspect of Age-Period-Cohort specificities in the US is connected to this pattern: did this topping of educational achievement have had an impact on other aspects of American’s social dynamics. In other terms: is this fluctuation in educational dynamics visible in other variables? which ones (such as social status, cultural habits, health conditions)?

 

5. third example: education and a comparison with Yang’s APC-IE

 

Here is an example of the limits of the Yang’s and colleagues apc_ie (intrinsic estimator) model. The aim of the ie is to provide a “per se” optimal age, period and cohort trend. When we make use of the apc_ie model for BA or higher owners, education seams to increase with period and to decrease with age, which is sociologically misleading.

www.louischauvel.org/apcdcpseduc.do

 

Figure5.1: shape of age, period and cohort coefficients of BA degree owners (or higher) in the US, after the Yang’s apc_ie model

 

 

In the general case, the linear trends of the age, period and cohort coefficient can not be interpreted easily and appear more as technical intercepts than as meaningful results. It is why we must prefer the detrended approach, where the real focus is driven to the shocks below or above the linear trends.