Announcement

Collapse
No announcement yet.
X
  • Filter
  • Time
  • Show
Clear All
new posts

  • Interaction terms to check if variables are interdependent?

    I have the potentially endogenous variables Loan size and the loan rate. Loan rate is my dependent variable and loan size is an Independent variable. I implemented an interaction of loan size with my test variables to see if my test variables are affected by loan size. Can this Count as an indirect check for endogeneity? As I could not find suitbale Instruments to conduct an instrumental variable Approach and dropping loan size turned my test variables insignificant, I want to check if endogeneity is a substantial problem in my model.
    input float Loanspread_pct int Age long Totalassets byte Numberofemployees float Corporationdummy long Grossprofit float(Profitability_pct Leverage_pct) long Loansize byte Maturity double(GDPGrowth Duration) byte Housebank str6 Loantype float Banks
    5.45 8 1500 28 1 1600 6.25 95 475 10 .015 0 0 "Credit" 7
    1.25 8 1500 28 1 1600 6.25 95 475 10 .015 0 0 "Credit" 1
    2.52 6 500 15 1 800 8.75 50 150 10 .015 5.75 1 "Credit" 8
    9.14 6 500 15 1 800 8.75 50 30 1 .015 5.75 1 "LC" 8
    9.07 6 500 15 1 800 8.75 50 20 1 .015 6 1 "LC" 8
    8.72 23 387 10 0 815 3.435583 72 80 1 .022 10 1 "LC" 8
    8.67 24 415 10 0 830 5.060241 77 80 1 .022 11 1 "LC" 8
    8.55 25 400 10 0 850 3.529412 90 120 1 .015 12 1 "LC" 8
    5.02 24 415 10 0 830 5.060241 77 60 6 .022 1 0 "Credit" 7
    9.42 15 800 25 1 3500 3.4285715 20 100 1 .015 4.666666666666667 0 "LC" 3
    4.07 15 800 25 1 3500 3.4285715 20 620 20 .015 0 0 "Credit" 6
    9.31 15 800 25 1 3500 3.4285715 20 230 3 .015 5 0 "LC" 5
    1.76 7 130 8 0 300 23.333334 40 50 10 .015 4.75 1 "Credit" 1
    .71 1 60 3 0 190 0 0 20 10 .005 0 1 "Credit" 1
    9.16 7 130 8 0 300 23.333334 40 15 3 .022 3 0 "LC" 3
    3.26 20 450 12 1 800 8.125 26 50 10 .015 10.083333333333334 0 "Credit" 8
    4.33 18 462 12 1 830 8.192771 32 125 5 .022 8 0 "Credit" 8
    5.17 19 438 12 1 755 7.549669 30 100 5 .022 0 0 "Credit" 4
    8.15 20 450 12 1 800 8.125 26 15 1 .015 10 0 "LC" 8
    8.26 19 438 12 1 755 7.549669 30 15 1 .022 9 0 "LC" 8
    8 18 462 12 1 830 8.192771 32 15 1 .022 8 0 "LC" 8
    8.84 19 438 12 1 755 7.549669 30 120 1 .022 10 0 "LC" 5
    8.97 18 462 12 1 830 8.192771 32 120 1 .022 9 0 "LC" 5
    8.67 20 450 12 1 800 8.125 26 10 1 .015 10.583333333333334 0 "LC" 5
    3.9 15 320 10 1 1000 8 55 70 6 .015 7 0 "Credit" 5
    4.09 15 320 10 1 1000 8 55 100 5 .015 5.166666666666667 0 "Credit" 4
    3.33 10 277 12 1 800 9.375 60 150 4 .015 5.083333333333333 1 "Credit" 5
    2.79 18 720 25 1 1800 11.38889 45 350 3 .022 12 1 "Credit" 5
    2.45 20 695 25 1 2000 10.5 45 300 6 .015 14 1 "Credit" 5
    4.55 3 248 3 1 500 11 44 30 4 .017 0 0 "Credit" 3
    4.91 4 250 3 1 600 8.333333 50 50 5 .022 1.33 0 "Credit" 4
    10.05 3 248 3 1 500 11 44 8 1 .017 0 0 "LC" 7
    10.02 4 250 3 1 600 8.333333 50 8 1 .022 1 0 "LC" 7
    10.03 4 250 3 1 600 8.333333 50 10 3 .022 1.083 0 "LC" 9
    9.84 2 462 25 1 1750 2.2857144 45 100 1 .022 0 0 "LC" 9
    9.43 3 450 29 1 1900 2.710526 50 200 3 .022 .5833333333333334 0 "LC" 9
    9.62 3 450 29 1 1900 2.710526 50 100 1 .022 1 0 "LC" 9
    5.26 2 462 25 1 1750 2.2857144 45 250 5 .022 0 0 "Credit" 4
    5.12 4 440 29 1 2000 2.5 50 200 5 .015 1.4166666666666667 0 "Credit" 4
    8.16 7 360 9 1 415 18.795181 25 15 1 .017 5 1 "LC" 7
    8.11 8 350 9 1 435 18.62069 25 25 1 .022 6 1 "LC" 7
    8.04 9 345 9 1 430 18.60465 30 15 1 .022 7 1 "LC" 7
    2.58 45 1000 14 0 1450 7.931035 60 350 7 .005 15 1 "Credit" 7
    2.27 50 1050 15 0 1500 6.666667 70 300 10 .015 20 1 "Credit" 7
    8.23 45 1000 14 0 1450 7.931035 60 150 1 .005 15 1 "LC" 7
    8.07 46 970 15 0 1400 6.785714 70 150 1 .022 16.5 1 "LC" 7
    8.05 47 960 15 0 1475 6.779661 70 150 1 .017 17.75 1 "LC" 7
    8.78 7 350 3 0 400 12.5 50 20 1 .015 7 1 "LC" 6
    3.39 7 350 3 0 400 12.5 50 15 5 .015 7 1 "Credit" 6
    2.9 25 500 25 1 1100 18.181818 80 150 10 .015 15 1 "Credit" 5
    2.6 25 500 25 1 1100 18.181818 80 400 15 .015 15 1 "Credit" 5
    8.52 25 500 25 1 1100 18.181818 80 50 1 .015 15 1 "LC" 5
    2.1 40 620 25 0 2000 15 20 150 10 .015 20 1 "Credit" 5
    8.12 40 620 25 0 2000 15 20 50 1 .015 20 1 "LC" 5
    2.57 35 380 12 1 1500 6.666667 30 25 5 .015 15 1 "Credit" 5
    3.12 4 400 7 0 950 13.68421 25 300 5 .017 3 1 "Credit" 7
    2.54 7 425 9 0 1000 12.3 20 250 7 .015 6 1 "Credit" 7
    8.76 4 400 7 0 950 13.68421 25 50 1 .017 3 1 "LC" 7
    8.82 5 415 8 0 975 14.358974 20 80 1 .022 4.333333333333333 1 "LC" 7
    8.87 6 410 9 0 935 13.368984 20 80 1 .022 5.333333333333333 1 "LC" 7
    8.66 7 425 9 0 1000 12.3 20 80 1 .015 6 1 "LC" 7
    2.85 102 370 6 0 427 14.285714 42 80 5 .022 23 1 "Credit" 5
    9.24 102 370 6 0 427 14.285714 42 30 1 .022 8 0 "LC" 6
    9.17 103 375 6 0 430 13.953488 45 45 1 .022 8.75 0 "LC" 6
    1.6 102 370 6 0 427 14.285714 42 80 5 .022 0 0 "Credit" 2
    1.91 17 3500 28 1 2875 5.495652 38 500 10 .005 14 1 "Credit" 5
    3.76 22 3625 30 1 3000 5 40 400 7 .022 4 0 "Credit" 3
    9.36 22 3625 30 1 3000 5 40 60 2 .022 5 0 "LC" 4
    9.68 22 3625 30 1 3000 5 40 50 2 .022 .16666666666666666 0 "LC" 4
    4.61 18 3100 15 1 2600 6.538462 50 150 3 .018 5 0 "Credit" 3
    4.65 18 3100 15 1 2600 6.538462 50 130 4 .018 4 0 "Credit" 4
    9.4 18 3100 15 1 2600 6.538462 50 50 2 .018 4 0 "LC" 4
    2.09 26 2650 35 1 2300 9 21 300 5 .022 22 1 "Credit" 7
    2.15 27 2710 35 1 2425 9.278351 28 250 7 .017 23 1 "Credit" 7
    1.98 29 2665 33 1 2400 8.75 25 50 9 .022 25.25 1 "Credit" 7
    1.75 30 2700 33 1 2350 8.297873 25 80 10 .015 26.333333333333332 1 "Credit" 7
    7.8 27 2710 34 1 2425 9.278351 28 80 1 .017 23.166666666666668 1 "LC" 7
    2.74 17 1980 26 1 1650 8.939394 26 325 10 .022 16 1 "Credit" 7
    2.7 19 2050 26 1 1700 8.941176 31 150 8 .022 18.333333333333332 1 "Credit" 7
    2.55 20 1930 26 1 1750 8.857142 33 220 5 .015 19.166666666666668 1 "Credit" 7
    8.15 19 2050 26 1 1700 8.941176 31 80 1 .022 18.166666666666668 1 "LC" 7
    end
    [/CODE]
    Code:
    xtset Banks
    Code:
    xtreg Loanspread_pct Age Totalassets Numberofemployees Corporationdummy Grossprofit Profitability_pct Leverage_pct Loansize Maturity GDPGrowth_pct Duration Housebank c.Loansize#c.Duration c.Loansize#i.Housebank if Loantype=="LC", fe vce (cluster Banks)

    Robust
    Loanspread~t Coef. Std. Err. t P>t [95% Conf. Interval]

    Age .0070297 .0067637 1.04 0.339 -.0095205 .0235798
    Totalassets .0008028 .0000978 8.21 0.000 .0005636 .001042
    Numberofem~s -.0151647 .0204591 -0.74 0.487 -.0652264 .034897
    Corporatio~y -.4132243 .2830996 -1.46 0.195 -1.105944 .2794954
    Grossprofit .0001536 .0002272 0.68 0.524 -.0004024 .0007096
    Profitabil~t -.0353282 .0320068 -1.10 0.312 -.113646 .0429896
    Leverage_pct .0144695 .0036746 3.94 0.008 .005478 .0234609
    Loansize -.0046372 .0026792 -1.73 0.134 -.011193 .0019186
    Maturity .00222 .1535463 0.01 0.989 -.3734942 .3779341
    GDPGrowth_~t -.0282088 .1915538 -0.15 0.888 -.496924 .4405065
    Duration -.141632 .0406531 -3.48 0.013 -.2411066 -.0421574
    Housebank .0503532 .5911262 0.09 0.935 -1.396081 1.496787

    c.Loansize#
    c.Duration .0000725 .0004771 0.15 0.884 -.0010949 .00124

    Housebank#
    c.Loansize
    1 -.0034348 .0105666 -0.33 0.756 -.0292903 .0224207

    _cons 9.61221 .5827605 16.49 0.000 8.186247 11.03817

    sigma_u 1.0533132
    sigma_e .28206771
    rho .93308652 (fraction of variance due to u_i)

    Also in respect of the Interpretation of the interaction coefficient, does the positive coefficient of Duration#Loansize mean that the effect of Duration on Loanspread_pct is stronger by
    .0000725 percentage points
    when loansize increases by 1000€?

  • #2
    Note that using a fixed spacing font makes your results posting more readable.

    "I implemented an interaction of loan size with my test variables to see if my test variables are affected by loan size. Can this Count as an indirect check for endogeneity?" - no, an interaction term is not a test for endogeneity. If you're working with banks, I would think that you have lots of earlier data to use for instruments.

    The easiest way to think through the interaction's effect is to use margins. Your interpretation seems OK assuming loan size in in 1000€.

    Comment


    • #3
      Thx for your Reply. I also included the interaction between my both test variables
      Code:
       xtreg Loanspread_pct Age Totalassets Numberofemployees Corporationdummy Grossprofit Profitability_pct Leverage_pct Loansize Maturity g1 g3 GDPGrowth_pct Duration Housebank c.Duration#i.Housebank if Loantype=="LC", fe vce (cluster Banks)
      --------------------------------------------------------------------------------------
      | Robust
      Loanspread_pct | Coef. Std. Err. t P>|t| [95% Conf. Interval]
      ---------------------+----------------------------------------------------------------
      Age | .0047148 .0032826 1.44 0.201 -.0033175 .0127471
      Totalassets | .0001963 .00032 0.61 0.562 -.0005868 .0009794
      Numberofemployees | -.0040658 .0153328 -0.27 0.800 -.0415839 .0334523
      Corporationdummy | -.4338111 .0941939 -4.61 0.004 -.6642954 -.2033268
      Grossprofit | -.0003241 .0001544 -2.10 0.081 -.0007018 .0000537
      Profitability_pct | -.0432664 .0300655 -1.44 0.200 -.116834 .0303012
      Leverage_pct | .0060082 .0042953 1.40 0.211 -.0045021 .0165185
      Loansize | -.0014868 .0014082 -1.06 0.332 -.0049326 .001959
      Maturity | -.0318797 .0938426 -0.34 0.746 -.2615043 .1977449
      g1 | -.4679261 .1417758 -3.30 0.016 -.8148391 -.1210131
      g3 | .7020761 .3979373 1.76 0.128 -.2716414 1.675794
      GDPGrowth_pct | -.0337004 .0386487 -0.87 0.417 -.1282703 .0608695
      Duration | -.1538444 .056374 -2.73 0.034 -.2917866 -.0159022
      Housebank | -.9398553 .1462606 -6.43 0.001 -1.297742 -.5819685
      |
      Housebank#c.Duration |
      1 | .0938528 .0360128 2.61 0.040 .0057327 .1819728
      |
      _cons | 10.77541 .5244268 20.55 0.000 9.492181 12.05863
      ---------------------+----------------------------------------------------------------
      sigma_u | .4269685
      sigma_e | .20457035
      rho | .81329994 (fraction of variance due to u_i)
      --------------------------------------------------------------------------------------
      I noticed that the negative coefficient of Duration almost doubled in size after i included the interaction term. Could the reason be that after including the interaction, the coefficient of Duration only reflects the case when housebank is Zero and has then a stronger decreasing effect on loanspread as withot the interaction when the coefficient of Duration reflects the weighted average of housebank being Zero and housebank being one?
      Since the coefficient of the interaction is positive and that of Duration negative, would Duration have a stronger decreasing effect when the housebank is 1 or would the positive sign equal out the negative sign of Duration?

      Comment

      Working...
      X